Uppsala University Master thesis in Accounting Department for Business Studies Auditing and Business Analysis, 30 Credits

Supervisor Mattias Hamberg Spring Semester 2013

I

UPPSALA UNIVERSITY · DEPARTMENT OF BUSINESS STUDIES · MAY 2013

Audit Fee Determinants in different

Ownership Structures

THE SWEDISH SETTING

Joakim Ask and Mattias LJ Holm

ABSTRACT

The aim of this study is to test the audit fee determinants for companies listed on Nasdaq OMX

Stockholm Stock Exchange and to examine whether the audit fee determinants diverge for

ownership structures. By testing the audit fee determinants in a Swedish setting the study

contributes to the research body in two ways; by testing a previously sparsely researched setting and

examining the monitoring need for different ownership structures. The results indicate that audit

fees are explained to a large extent by accounting complexities, business complexities and

assurance. The results also suggest that ownership structure does not have a large effect on the

monitoring need. Altogether the results provide further evidence on audit fee determinants whilst

adding initial insight into the area of audit fee determinants for ownership structures.

Keywords · Accounting complexity, Agency theory, Assurance, Auditing, Audit fee determinants,

Business complexity, Monitoring cost, Ownership structure

II

ACKNOWLEDGEMENT

First and foremost we would like to thank our supervisor Mattias Hamberg, who has been

outright most helpful during the whole process of writing this thesis by providing

knowledgeable tips and inspiration. We would also like to thank our seminar group for helpful

comments and discussions.

We would also like to thank the IASIP gang for providing comedic relief during many

stressful periods.

III

TABLE OF CONTENTS

1 INTRODUCTION ................................................................................................................. 1

1.1 Auditing and audit fees ..................................................................................................... 1

1.2 Audit fee determinants ...................................................................................................... 2

1.3 Research purpose .............................................................................................................. 4

1.4 Thesis disposition ............................................................................................................. 4

2. THEORETICAL FRAMEWORK ..................................................................................... 5

2.1 Agency theory ................................................................................................................... 5

2.2 Monitoring mechanism and costs ..................................................................................... 6

2.3 Audit fee determinants ...................................................................................................... 7

2.3.1 Accounting complexities ............................................................................................ 9

2.3.2 Business complexities ............................................................................................... 12

2.3.3 Assurance ................................................................................................................. 17

2.4 Synopsis .......................................................................................................................... 20

2.5 Hypotheses ...................................................................................................................... 20

3. RESEARCH METHOD .................................................................................................... 22

3.1 Overall research design .................................................................................................. 22

3.2 Sample selection and data sources .................................................................................. 22

3.3 Measures ......................................................................................................................... 25

3.3.1 Dependent variable .................................................................................................. 25

3.3.2 Independent variables .............................................................................................. 25

3.4 Audit fee model specification ......................................................................................... 29

3.5 Statistical tests ................................................................................................................ 29

4. RESULTS ............................................................................................................................ 31

4.1 Descriptive statistics ....................................................................................................... 31

4.2 Correlations .................................................................................................................... 32

4.2.1 Interaction between the independent variables ....................................................... 33

4.3 Regressions ..................................................................................................................... 34

4.3.1 Primary audit fee model ........................................................................................... 35

4.3.2 Secondary audit fee model ....................................................................................... 41

4.4 Robustness tests .............................................................................................................. 44

5. CONCLUDING REMARKS ............................................................................................. 47

5.1 Conclusions .................................................................................................................... 47

IV

5.2 Possible limitations ......................................................................................................... 48

5.3 Suggestions for further studies ....................................................................................... 49

REFERENCES .......................................................................................................................... I

APPENDIX ........................................................................................................................... VII

TABLE OF FIGURES

1 Synopsis of previous research and expected relationships .............................................. 20

2 Sample description ............................................................................................................. 23

3 Presentation of the variables, estimations and variable measurement .......................... 28

4 Decscriptive statistics .......................................................................................................... 31

5 Correlations amongst the variables in the audit fee model ............................................. 32

6 Primary regression model and bivariate test for audit fee determinants ...................... 35

7 Secondary regression model for audit fee determinants ................................................. 42

8 Primary regression model modifed and tested per audit fee determinants group ........ 44

1

1 INTRODUCTION

1.1 Auditing and audit fees

Why do we need auditing? According to Wallace 1980 the need for auditing is threefold; to

function as a monitoring mechanism, to meet the demand for information from investors in

order to improve their decision-making and to act as an insurance against material

misstatements. Pentland (2000) described the explosion in the demand for auditing as a

control mechanism which has been even further accentuated during the 21st century. The

demand for auditing as a control mechanism has increased even further as the world have

been rattled by economic crises and scandals sparking a plea for stricter control and regulation

of companies. Auditing can be viewed as a corporate governance function with the purpose of

creating confidence in the financial statements issued towards stakeholders (Pentland, 1993)

and to enhance the “credibility of the audited object” (Power, 1999, p.28). Furthermore Schilt

& Perler (2010) states that auditing has an important role within the financial system, both as

an assurance service and as a way to protect investors from opportunistic behaviour. Instances

where the auditor has failed to protect the investors and provide assurance has led to a

lessened trust in auditing as a profession. For example, the appointed audit firm in the Enron

scandal (Arthur Andersen) was disbanded as investors had lost faith in Arthur Andersen’s

independence and capabilities (Schilt & Perler, 2010).

Auditing is often blamed in the wake of large scandals, leading to a stricter regulation with the

purpose of increased audit quality (Francis, 2004). Regulation of audit and accounting

services, for example IFRS and ISA1, has tended to put an upward pressure on the pricing of

audits as well as increasing the complexity of auditing (Kim, Liu & Zheng, 2012; Taylor &

Simon, 1999). It has also been suggested that increased regulation can affect the price of

different goods and services of the audited company (Taylor & Simon, 1999). Francis &

Wang (2005) investigated whether disclosure of the audit fees would lead to a more precise

audit pricing, their study shows that clients that were overcharged (undercharged) before the

introduction had lower (higher) audit fees and total fees the subsequent year after introduction

(Francis & Wang, 2005). These findings suggest that the introduction of audit fee disclosure

had a positive effect on the accuracy of audit pricing.

1 IFRS is an abbreviation for International Financial Reporting Standards, ISA is an abbreviation for International Standard on Auditing

2

Sweden was fairly early to introduce regulation demanding disclosure of audit and non-audit

fees. The regulation is vested in the 5th

Chap. 9 § of the Swedish Annual Accounts Act

(Ministry of Justice, 1995) and states that major limited companies shall disclose the total

accrued remunerations paid for the fiscal year to each of the auditors as well as the total

remunerations paid to the company where the auditors are employed. The companies shall

also disclose information regarding how the accrued remunerations have been distributed

amongst audit service, near-audit service, tax advisory and other services.

Power (2003, p.392) describes the audit practice as “a self-regulating system whose

components are an interacting, semi-institutionalized and loosely coupled whole, a structure

that is constantly moving and subject to economic, regulatory and political pressures for

change”. Economic, regulatory and political pressures can be challenging for an auditor as the

auditor tries to find a balance between the core professional values of the auditing profession,

i.e. objectivity and independence (Suddaby, Gendron & Lam, 2009). This might suggest that

in order to maintain the objectivity and independence an auditor has to be separated from

commercial pressures. The auditor also has to devote an optimal amount of time for the audit

assignment with a selected qualified staff in order to achieve a high audit quality (Lee, 2002).

Due to cost concerns it is nevertheless seldom in the audit client’s interest to have the auditor

perform a full audit where everything is scrutinized. Higher costs of the service performed

(e.g. audit) are normally associated with higher quality (Sundgren & Svanström, 2013) but as

the costs of a complete audit tends to exceed the benefits of a complete audit (Francis, 1994)

they are not most often not desirable from a client’s perspective.

1.2 Audit fee determinants

Audit fees and the determinants of audit fees have historically garnered quite a large amount

of research interest in the US; the research area have however been sparsely studied within the

European setting. Important research on audit fee determinants in the US market was

conducted in the early 1980’s by Dan A. Simunic. In a study on price competition in the audit

market, Simunic (1980) developed a model to determine which factors determine audit fees.

The basic assumption is that audit fee differences can be twofold, either due to disparities in

time spent on an audit or due to differences in the charged price (Simunic, 1980). The model

has since then been deployed in multiple audit fee studies with variations in the tested audit

fee determinants (Cobbin, 2002; Hay, Knechel & Wong, 2006). The increasing globalization

3

of the world economy does raise the need for research on audit fee determinants in Europe

due to the idiosyncrasies of the European economy (Cobbin, 2002).

Previous research has exposed variances in the audit fee determinants for different

geographical regions (Cobbin, 2002). The audit fee determinants in Scandinavia, which share

some common characteristics, were first tested in Norway. At the time of the study on the

Norwegian audit fee determinants there were two audit firms which had a stronghold on the

audit market (Firth, 1997). A market stronghold would indicate that the two firms should be

able to charge higher audit fees than the rest of the competitors, the study did however not

find any proof that the two firms were able to transform their stronghold into higher charged

audit fees (Firth, 1997). There are numerous amounts of different country-specific audit fee

determinants which need to be tested in order to reach consistency regarding audit fee

determinants globally.

Hay et al. (2006) used a meta-analysis approach to test whether the most commonly used

audit fee determinants within the research body collectively can reach unanimity regarding

which are the most important audit fee determinants. In consensus with other research Hay et

al. (2006) conclude that there are some common main determinants of audit fees, namely the

size of the firm, the complexity of the firm and the inherent risk of the firm. Earlier research

has also recognized a large number of potential audit fee determinants ranging from merger &

acquisitions (Menon & Williams, 2001), whether the auditor is from a Big 42 audit firm

(Simunic, 1980; Taylor & Simon, 1999; Waresul Karim & Moizer, 1996) to ownership

structure (Nikkinen & Sahlström, 2004).

A concentrated ownership can lead to agency problems; the tight control that is implied by a

high owner concentration creates a situation where self-serving behaviour can go on

unchallenged. The controlling owner can exploit the power that comes with a large ownership

to serve their own best interest. In a study by Fan & Wong (2005) on the external auditor’s

ability to function as corporate governance tool, the ownership structures effect on agency

theory is investigated. They states that as the percentage of voting rights increase for a

controlling owner the more powerful the controlling owner’s position within the company

becomes and therefore an agency problem arises. To mitigate the effects of the agency

problem the controlling owner can implement a credible monitoring mechanism, e.g. auditing.

2 Big 4 have in previous research been denoted Big 8/7/6/5 due to the then current audit environments. The Big 4 firms are Ernst & Young, Deloitte, KPMG and PWC.

4

The power structure in a company with a highly concentrated ownership can result in a

different monitoring mechanism role then for a company with a less concentrated ownership.

From an auditing perspective this could influence the audit fee determinants.

1.3 Research purpose

The purpose of this study is to test which factors determine the audit fee and whether the

ownership structure of a company increases the importance of these audit fee determinants.

1.4 Thesis disposition

The study continues as followed, in Chapter 2 relevant theories that frame the research area

are discussed, audit fees are defined as a monitoring cost in the agent – principal relationship.

The monitoring problem areas are then used to identify the audit fee determinants which are

presented and divided in to three main categories; accounting complexity, business

complexity and assurance. The chapter concludes with a hypothesis development. Chapter 3

describes the overall research design and the data utilized to test the different determinants

relationship to audit fees. Furthermore, the measures for the independent and dependent

variables are presented and a model for testing is constructed. Chapter 4 presents the

descriptive statistics for the variable data used in the thesis, the results from the Pearson

correlation test and the results from the primary and secondary regression models. The results

are analysed by discussing the interactions between the independent variables and the possible

problem regarding multicollinearity. The chapter continues with the analysis of the two audit

fee models presented in chapter 3, where the primary model is discussed through the

theoretical framework. The secondary model adjusted for ownership structure is then analysed

and the effects of having single owner with at least 50 % of the voting rights on audit fees is

discussed. Robustness tests conclude the chapter. Chapter 5 concludes the thesis by

presenting the conclusions and limitations of the thesis and offers suggestions for further

research on the area of audit fee determinants.

5

2 THEORETICAL FRAMEWORK

The first sections (2.1 - 2.2) of this chapter defines audit fees as a monitoring cost that is

derived from the agent - principal relationship that exists between the shareholder(s) and

manager(s) of an organization. The following section 2.3 reviews previous research and

presents the problematic monitoring areas and their estimated relationships to audit fees.

Summarizing the chapter is a synopsis of previous research and the estimated relationships

which works as a foundation for the hypothesis development.

2.1 Agency theory

An agency relationship can be defined as “a contract under which one or more persons (the

principal(s)) engage another person (the agent) to perform some service on their behalf which

involves delegating some decision making authority to the agent” (Jensen & Meckling, 1976,

p.308). There a many examples on different agency relationships or as Ross (1973, p.134)

states “examples of agency are universal”. A classic agency relationship is the management –

shareholder relationship in a company where the management, being the agent, is expected to

act in the best interest of the principal, i.e. the shareholders. More or less all relationships

governed by a contract do nevertheless contain some elements of an agency relationship

(Ross, 1973).

The relationship can be problematic though, when both parties are expected to act rationally

whilst trying to maximize their own utility which might not always be perfectly aligned

between the two parties (Ross, 1973). The formal expectation of an agency relationship is that

the agent acts with the best interest of the principal at heart whilst having the principals

consent to do so (Mitnick, 1973). In reality however, what benefits the agent might not benefit

the principal leading to a conflict of interest, Mitnick (1973) describes this as the principal’s

problem and that the principal must “motivate the agent to act for the principal’s goal in the

manner the principal prefers”. The question that arises is; how can the principal ensure that

the agent acts in the best interest of the principal? The principal needs to assure that his or her

best interest is being adhered to through some kind of monitoring of the agent which can be

quite troublesome (Ross, 1973).

In excess of the sheer problems with monitoring their might be instances where it is also not

economically feasible to monitor the agent (Ross, 1973) which could indicate that a situation

where an information asymmetry between the agent and the principal could arise. In a study

6

on CEO return and shareholder interest, Nyberg, Fulmer, Gerhart & Carpenter (2010) states

that the lack of common goals, interest and asymmetry in available information leads to an

increased need for monitoring as well as incentives for managers to behave in a desired way.

Monitoring and incentive packages results in agency costs, i.e. the cost for controlling the

agent’s behaviour (Nyberg et al., 2010).

Stakeholders often lack the opportunity to partake in a company’s everyday business and

instead they employ managers who then are responsible for the decision-making in the

company. The employment of managers’ leads to a need for information regarding the actions

the managers undertakes, Gjesdal (1981) defines this as stewardship demand. While agency

theory assumes that both parties are utility maximizing in the short term, stewardship theory

assumes that the agent have a moral obligation towards the organization and its shareholders

(Hernandez, 2012). Hence, the agent and the shareholders’ interests will tend to align more

and in order to maximize organizational performance the agent will act with the best interest

of the organization at heart (Fox & Hamilton, 1994) and thereby foregoing the individual self-

interest. In accounting terms, the existence of stewardship theory would lead to managers

actually publishing information about their use of resources to those they are accountable

without the need of monitoring (Gjesdal, 1981). Therefore, in organizations where

stewardship theory holds agency costs would be greatly reduced. In terms of auditing this

would mean, given that stewardship theory holds, that audit fees would be lower in

organizations than if agency theory would hold.

2.2 Monitoring mechanism and cost

In cases where both parties in the agency relationship are utility maximizing there is reason to

believe that the parties’ interests will differ (Jensen & Meckling, 1976). To ensure that the

agent acts in the best interest of the principal(s) incentives and/or monitoring is applied by the

principal at a cost, the agency cost (Jensen & Meckling, 1976). According to Jensen &

Meckling (1976) the agency cost consist of three parts; the monitoring expenditures by the

principal, the bonding expenditures by the agent and the residual loss. One important part of

the monitoring expenditures is the audit fee, “since auditors have a duty to ensure that the

managers are behaving according to the owners’ interest while they also have a duty to

inspect the company’s accounts” (Nikkinen & Sahlström, 2004, p.254). Therefore, the auditor

can be seen as the guardian of the organization who is monitoring its behaviour with the

organizations formal goals as guidance (Mitnick, 1973).

7

All listed companies in Sweden are required to appoint an auditor3 during the annual general

meeting (SFS 2005:551, 9th

Ch., 8 §). The auditor’s role according to the ISA is to enhance

the confidence in the financial statements to its intended users by obtaining reasonable

assurance that the financial statement is free from material misstatements whether due to

fraud or error (IFAC, 2009). Through an agency theory perspective, the auditor is the

monitoring mechanism implemented to ensure that the managers act in the best interest of the

owners and also to inspect the company´s accounts (Nikkinen & Sahlström, 2004). The

appointment of an auditor as a monitoring mechanism will result in a monitoring cost, the

audit fee. The audit fee is determined by two main factors, the audit pricing per unit and the

amount of time spent on the audit (Simunic, 1980).

However, if the agents are numerous the only viable communications mechanism might be

the monitoring cost and “while it might be feasible to monitor the agent´s actions, it would

not be economically viable to do so” (Ross, 1973, p.138). This implies that in cases where

there are numerous agents the agents’ are controlled by the bonding cost which prohibits the

agent from taking certain actions or compensates the principal if the agent does undertake

prohibited actions (Jensen & Meckling, 1976). The agency costs are at its highest when the

principal and the agent’s interest significantly diverge, i.e. the monitoring and bonding

expenditures are at its highest when there is a high degree of separation of ownership and

control (Fleming, Heaney & McCosker, 2005). Agency costs are at its lowest when the

parties’ interests are aligned and only the residual cost applies (Jensen & Meckling, 1976). In

financial economy, the agency cost is at its lowest in a 100% owner-managed firm because

the shareholders and the managers’ interests align (Fleming et al., 2005).

2.3 Audit fee determinants

The audit fee determinants applied in this study are divided into three groups; Accounting

complexities, business complexities and assurance. Dividing the variables into these three

groups will improve the study as the results generated from the empirical study can be

analysed according to its underlying characteristic. The following sections (2.3.1 – 2.3.3 have

a focus on the previous literature on audit fee determinants and the section is relatively

measurement oriented.

3 If more than one of the following demands are fulfilled; the company has more than three employees, assets over 1,5 MSEK and a net

turnover above 3 MSEK (SFS 2005:551, Ch. 9, 1§)

8

The accounting complexities is related to the problematic areas of the financial accounts,

Peterson (2012, p.5) defines accounting complexity as “the amount of uncertainty related to

the mapping of transactions or potential transactions and standards into the financial

statements”. Some accounting areas are rather straightforward while other areas are more

complex and can lead to subjective choices affecting the comparability of financial

statements. The subjective nature of some accounting areas can lead to misstatements, either

intentional or by mistake, increasing the information asymmetry between managers and

shareholders. An information asymmetry is problematic from an agency theory perspective

(Ross, 1973) as an information asymmetry increases the principal’s need for monitoring of the

agent’s activities (Nyberg, et al., 2010).

Business complexities are related to the inherent complexities arising from the nature of a

company’s business. According to Esteves (2006) the methods for measuring business

complexity can be based on size, diversification and divisionalization of the organization. A

company´s business complexity can be considered as the “organizational structure and

inherent complexities of their business activities” (Esteves, 2006, p. 2). The business

complexities of a company increase the demands on the auditor. As the agent undertakes

complex business decisions the best interest of the agent and principal might not align (Ross,

1973) which is the reason for the increased demand on the auditor. These complex business

decisions can damage the principals, the risk for damage increases the need for monitoring i.e.

auditing.

Assurance is related to areas where trust in the auditors work is increased through the

implementation of various approaches. The auditor should provide a high level of assurance

to the stakeholders ensuring that the financial statement is free from material misstatements

(European commission, 2010). However, ISA 200 (2009) states that an auditor is not required

to give an absolute assurance that the financial statements are free from material

misstatement. Instead, the auditor should provide reasonable assurance (IFAC, 2009), which

is a high level of assurance that the financial statements as a whole are free from material

misstatements. Reasonable assurance is reached “when the auditor has obtained sufficient

appropriate audit evidence to reduce audit risk (that is, the risk that the auditor expresses an

inappropriate opinion when the financial statements are materially misstated) to an

acceptably low level” (ISA 200, 2009, p.74). The variables in the group assurance are tools

9

used to limit the risk of material misstatement and thereby increase the assurance.

2.3.1 Accounting complexities

Proportion of financial assets

The demand for auditing has had a steep upward incline in the recent decade as stakeholders

demand accountability to protect their interests (Pentland, 2000). This accountability is

produced through the verification of financial accounts, a process that is being conducted by

the auditors. Pentland (1993) describes the process as a ritual where the auditors reach a

certain level of comfort throughout the audit team, this level of comfort then reaches the

signing partner who then verifies the accounts which in turn generates comfort to the

stakeholders. Furthermore, verifiability is a key component for the auditability of asset

valuation (Power, 1999), “there is a close link between the credibility of economic

measurement for accounting recognition purpose and the auditability of such measures”

(Power, 1999, p.79).

The auditability of the measures in the verification process are however not the same for all

assets. Brand valuations can be quite troublesome as an exact measurement might be hard to

derive and the verification is more dependent on judgment then calculations (Power, 1999)

which is in contrast to the verification of financial assets, such as stocks, where the focus is on

the calculations. Therefore, the estimation is that companies with a large proportion of

financial assets to total assets will provide an easier verification process for the auditors and

thereby demand less time of the audit which reduces the audit fee.

Impairment of intangible assets

Under the standard IAS 36 (IASB, 2004) a company have to perform annual impairment tests

for goodwill and certain other intangible assets, as well as conduct impairments tests for other

assets when there might be reason to suspect value impairment. The aim of the standard is to

“prescribe the procedures that an entity applies to ensure that its assets are carried at no

more than their recoverable amount” (IASB, 2004, p.12). However, application of the

standard can be quite problematic, with opponents of the standard claiming that the

application of the standard leads to unverifiable accounts with low reliability (Bens, Heltzer &

Segal, 2011). Consequently, auditing the impairments might be troublesome in a number of

ways. When auditing a troublesome account the auditor either have to devote more time than

10

usual or employ auditors with expertise knowledge regarding the specific account. Expertise

knowledge does tend to have a higher audit pricing and the extra time spent on the audit can

lead to a higher charged audit fee which suggests a positive relationship between impairments

of intangible assets and audit fees.

Accruals

Accruals can be problematic from an auditing perspective; managers can exercise some

degree of discretion surrounding the accruals and that discretion can be used to influence the

future expenses and revenues (Srinidhi & Gul, 2007). Mistakes in accruals can also be made

even if the manager does not have any malicious intent. Hence, accruals could be an

accounting area where the risk of material misstatements is higher than an auditor can accept.

To reduce the risk of material misstatements in the accruals accounts the auditor has to

increase the audit effort and utilize their competence to asses if the accrual estimations are

correct (Srinidhi & Gul, 2007).

Previous research on the area of audit fees and accruals is sparse. Srinidhi & Gul (2007)

studied the effects of audit and non-audit fees on accrual quality and found that there is a

positive relationship between audit fees and accrual quality. Srinidhi & Gul (2007) findings

indicate that when the auditor devotes more effort to the audit assignment the accrual quality

is higher due to the limiting of managers opportunistic behaviour through accruals. Hence,

higher total accruals should require an increased audit effort to limit the effects of

opportunistic behaviour. Therefore a positive relationship is expected.

Provisions

IAS 37 defines provisions as “a liability of uncertain timing or amount”, and for a provision

to be recognized the following three criteria has to be meet (IASB, 2001):

A present obligation (legal or constructive) has arisen as a result of a past event,

payment is probable (more likely than not), and

The amount can be estimated reliably.

When discussing the amount of the provision IAS 37 uses the term “best estimate” of the

obligation at balance sheet date, and differentiates between one-off events (restructuring,

environmental and lawsuits) and large population events (warranties, customer refunds). One-

off events is measured at the most likely amount while large population evens are measured at

11

expected value (IASB, 2001). It is in the estimation of one-off events that management is

given the possibility for subjective estimations, which could lead to earnings smoothing.

Managers may have incentives for earnings smoothing to beat several targets (Moehrle,

2002), to increase future income (Bens & Johnston, 2009), and for anticipatory big bath

accounting (Peek, 2004). The expectation is that large provisions would increase the time

auditors have to spend on monitoring the provisions accounts and therefore a positive

relationship exists between the change in provisions and audit fees.

Free cash flow

Agency theory, which analyses the conflicts in the corporate managers (agents) and

shareholders (principals) relationship, suggests that conflicts of interest are severe when there

is a substantial generation of free cash flow within the organization (Jensen, 1986). The

notion is that large cash reserves increases the number of available options for corporate

managers to use the funds (Nikkinen & Sahlström, 2004). Free cash flow is “cash flow in

excess of that required to fund all projects that have positive net present value when

discounted at the relevant cost of capital” (Jensen, 1986, p.2). The problem arising when the

amount of free cash flow increases is how to assure that the manager uses the free cash flow

in the best interest of the shareholders rather than “investing it at below the cost of capital or

wasting it at organizational inefficiencies” (Jensen, 1986, p.2).

Nikkinen & Sahlström (2004) links free cash flow to audit fees as the auditors should be used

to monitoring the behaviour of the managers. Furthermore, Gul, Tsui and Chen (1998) and

Nikkinen & Sahlström (2004) both hypothesize that the relationship between free cash flow

and audit fees are positive. Nikkinen and Sahlström (2004) findings asserts that a positive

relationship between audit fees and free cash flow exists in a number of countries and that

agency theory can, to some extent, explain audit fees internationally. Based on Gul et al.

(1998) and Nikkinen and Sahlström (2004) this study estimates the relationship between audit

fees and free cash flow to be positive.

12

Inherent risk

Inherent risk is defined as the risk that a material misstatement is made in a separate account

or a number of accounts in the financial statements due to the accounts specific error of risk

(Maletta, 1993). Areas that are more difficult to audit heightens the demand for specific audit

procedures performed by experts (Simunic, 1980) which can lead to an increase in both time

spent on the audit as well as in the audit price per unit (Zerni, 2012). In low inherent risk

settings the auditor can depend on the company’s internal control function to a further degree

than in a situation where the inherent risk is high. When the inherent risk is high an auditor

has to employ both more complex and time-consuming audit tools (Maletta, 1993) since the

trust the internal control provides is not sufficient enough to protect against material

misstatements.

As the risk of material misstatements increases, so does the risk of future litigation and it is

therefore in the auditor’s interest to limit the risk of misstatements by dedicating more time to

the audit assignment. Previous research have indicated that the inherent risk of the company

leads to higher audit fees as the auditor tries to safeguard against future litigation (Taylor &

Simon, 1999) and as the need for specialists increases (Hay et al., 2006; Simunic, 1980; Stice,

1991). Based on the demand for more time dedicated to the audit and the need for expertise a

positive relationship between the audit fees and the inherent risk is expected.

2.3.2 Business complexities

Media exposure

The exposure in media for a company might have an effect on their corporate reputation

which in turn can influence the audit fee. Earlier research has confirmed an existence of a

positive relationship between corporate reputation and financial performance and the

corporate reputations potential for value creation (Roberts & Dowling, 2002). Furthermore,

Roberts & Dowling (2002, p.1077) recognize corporate reputation as an intangible asset that

“are critical because of their potential for value creation, but also because their intangible

character makes replication by competing firms considerably more difficult”.

Roberts & Dowlings (2002) research reveals that companies with a relatively good reputation

has a greater chance to sustain positive performance over time which should qualify the

corporate reputation as a factor to consider for auditors when performing the “going concern”

assessment of the audit client. Current corporate reputation of a company is decided by the

13

signals that reaches the public from either the company itself or other information channels

(media, stock market, etc.) and depicts the company’s behaviour and events (Brammer &

Millington, 2005; Brammer & Pavelin, 2006).

Furthermore Brammer and Pavelin (2006, p.34) states that “media exposure increases the

stock of available information, and also sets the context within which it is viewed” and

therefore play an important role in influencing corporate reputation and thereby the

performance overtime. This study will test if the news media coverage of a company does

have an influence on the audit fees. The expectation is that high visibility increases the

interest for the company. An increased interest can increase the litigation risk due to more

scrutiny and thereby more time should be spent on the audit assignment which could lead to a

higher audit fee.

Mergers and Acquisitions

Mergers and acquisitions (M&A) are generally described as a complex phenomenon

(Cartwright & Schoenberg, 2006; Meglio & Risberg, 2010) with inconclusive motives

(Nguyen, Yung & Sun, 2012) and a high failure rate (Angwin, 2007; Craninckx &

Huyghebaert, 2011). One reason for the confusion regarding the leading motives for an M&A

are according to Angwin (2007, p.101) the linkage between “abstract categories, and

performance rather than real motives and intended performance”. The motivational factors

exist only on a theoretical level and are not cross-referenced with the result generated from

the M&A.

However in an ex-post market evidence study Nguyen et al. (2012) find that most M&A have

multiple motivational factors with a combination of market timing, agency/hubris and

response to industry/economic shocks. These motives can be either value increasing or value

decreasing (Nguyen et al., 2012). The value increasing motives can be the benefits derived

from combining two companies and the value decreasing motives can be managers’ personal

interest for growth with disregard to the value for the shareholders. But whatever the

motivational factor behind the M&A might be, the fact remains that most M&A fail (Angwin,

2007; Craninckx & Huyghebaert, 2011).

M&A has been used in numerous cash flow and periodization manipulation scandals by

shifting cash outflows from the operating section to the investing section, or shifting cash

inflows from the investing section to the operating section of the cash flow statement (Schilt

14

& Perler, 2010). The amount of M&A is expected to increase the hours spent on the audit

assignment and therefore we expect M&A to have a positive relationship with audit fees.

Executive compensation programs

Diversity in the agent- principal relationship may occur when the two parties strive towards

different goals, to ensure that the principal acts in the best interest of the agent incentive

programs are often introduced (Nyberg et al., 2010). These incentives can either consist of

compensation to the agent when the company reaches predetermined goals and key figures or

in the form of equity stake offered to the CEO (Nyberg et al., 2010). The effectiveness of

these incentive programs are according to Chng, Rodgers, Shih & Song (2012) dependent on

the CEOs core self-evaluation and firm performance. When the firm performance is low

CEOs with a higher core self-evaluation will respond better to the challenges an incentive

program presents than CEOs with a lower core-self-evaluation (Chng et al., 2012). However,

many incentive programs tends to promote aggressive decision making by the manager, for

examples incentive programs based on growth can lead to ill-considered acquisitions to reach

the growth target.

Stock option based incentive programs can reduce the aggressive behaviour as the risk of

declining firm performance resulting in sufficiently lower profits will render the incentive

useless to the CEO thereby promoting less risk taking (Reitman, 1993). Declining firm

performance can however shift the CEOs focus to other performance based incentives

(Reitman, 1993). A possible problem with these incentive programs is the risk of

opportunistic behaviour by the CEO in order to reach the incentive programs goal resulting in

misstatements in the financial reports. Morse, Nanda & Seru (2011, p.1817) found that

“rigging explains between 10% and 30% of the incentive pay sensitivity to performance” and

that rigging increases with firm uncertainty but can be reduced by stronger governance. A

positive relationship between audit fees and the existence of executive compensation

programs is expected as the existence of executive compensation programs increases the

monitoring need.

15

Size of the audited firm

Due to its sheer size, a large-sized company requires the auditor to spend more time on the

audit assignment to fulfil the audit objective. The auditor have to devote more time on

maintaining the client-relationship as well as testing, both in detail and in a more holistic

setting, and understanding the clients specific business- and accounting- problems as the

company size increases. This indicates that the large-sized company will purchase more

auditing services compared to a small-sized company and thereby the audit fee should be

higher for a large-sized company (Palmrose, 1986a).

In the large-sized company segment the Big 4 auditor appointment is significantly more

common than in the small-sized company segment (Simunic, 1980). Big 4 auditor tend to

receive an audit fee premium (Choi, Kim, Liu & Simunic, 2008; Ebrahim, 2010) due to the

belief that Big 4 auditors provide a higher level of quality. Therefore, as the number of large-

sized companies appointing Big 4 auditors is higher than the number of small-sized

companies doing so the audit fee should increase with company size due to the Big 4 audit fee

premium.

The relationship between audit fees and company size is well established in academia (Menon

& Williams, 2011). Previous research has extensively tested the relationship between audit

fees and firm size and have found that a positive relationship exists which suggests that as a

company grows the audit fees increases (Firth, 1997; Simon & Francis, 1988; Simon, 1985;

Simunic, 1980; Taylor & Simon, 1999; Waresul Karim & Moizer, 1996). In line with

previous research we expect to find a positive relationship between audit fees and the audited

company’s size.

Number of employees situated abroad

The complexity of the audited company can vary based on a large wide array of factors, for

example it would seem reasonable to expect that a company which is spread extensively

geographically would be more difficult to audit than a company located within a smaller

geographic area. As the number of subsidiaries in a company increases, the amount of time

required to perform the audit increases (Menon & Williams, 2001) which would indicate that

as the geographical spread increases the complexity also increases. The complexity of a

client’s financial statement will increase the risk of litigation due to the heightened risk of

material misstatements (Taylor & Simon, 1999). Due to the heightened risk of litigation the

16

auditor should devote more time to testing transactions and examining the different accounts

in the financial statement. Hence, the more complex a client’s business is the more time has to

be devoted to the audit in order to ensure that the audit is of sufficient quality (Simunic,

1980).

As the auditor has to spend more time testing and controlling the financial statements the

audit fees should increase; previous research have generally found a positive relationship

between the audit fees and the client’s complexity by using different proxies of what

constitutes complexity (Menon & Williams, 2001; Simon & Francis, 1988; Simunic, 1980;

Taylor & Simon, 1999; Waresul Karim & Moizer, 1996). However, the proxies for

complexity have varied quite a lot in previous research; the most commonly used proxy has

been the number of subsidiaries followed by the number of foreign subsidiaries (Hay et al.,

2006). Based on previous research a positive relationship between foreign operations and

audit fees is expected.

Ownership structure

The monitoring costs is generally associated with the separation of ownership and control

(Fleming et al., 2005; Jensen & Meckling, 1976) and are assumed to be the lowest for

companies with 100% owner-managed firms (Fleming et al., 2005). Fleming et al. (2005) find

that a reduced separation of ownership and control generally leads to lower monitoring costs,

“agency costs are lower in firms managed by equity holders” (Fleming et al., 2005, p.50).

The managerial ownership has been found to be a significant factor influencing monitoring

costs in a number of western studies (Fleming et al., 2005; Jensen & Meckling, 1976;

Nikkinen & Sahlström, 2004). Furthermore, in a study by Mustapha & Ahmad (2011) on the

effects of managerial ownership in Malaysia and its relation to agency theory find evidence

for the relationship between manager ownership and monitoring costs in a non-western setting

suggesting that the relationship may hold despite cultural factors.

The monitoring of the agent which in this case is the corporate managers is conducted by the

auditors where the audit fee is the product of that monitoring (Nikkinen & Sahlström, 2004).

Nikkinen and Sahlström (2004) confirm the existence of a negative relationship between audit

fees and ownership structure. The principals and agents interests are more likely to coincide

when the manager(s) own a large proportion of the shares. Concentrated ownership can lead

to self-serving behaviour as the controlling owner has more power within the company. Fan &

17

Wong (2005) suggests that to limit the agency problem arising from the power structure the

controlling owner can employ monitoring mechanisms to increase his or hers credibility. The

total audit fee should however be lower for a company with a concentrated ownership as the

best interests are more likely to coincide and as the risk for information asymmetry decreases.

Based on Nikkinen & Sahlström (2004) results a negative relationship between audit fees and

ownership structure is expected.

2.3.3 Assurance

Non-audit fees

Early research on the connection between audit fees and non-audit services (NAS) identified a

positive relationship (Firth, 1997; Palmrose, 1986b; Simon, 1985; Simunic, 1980). In the

important research on audit fee determinants by Simunic (1980) a positive relationship

between audit fees and NAS was found. Simunic (1980) research was then replicated and

tested by Simon (1985) to examine whether the results can be generalized for other years than

the year Simunic tested. The results were found to be robust and a positive relationship

between audit fees and NAS could be established. Palmrose (1986b) found further evidence

for the positive relationship to audit fees for three categories of NAS; accounting-, non-

accounting and tax related with the strongest relationship being between audit fees and

accounting related NAS.

However, later research on the audit fee/NAS relationship has garnered other results. Firth

(1997) explored the possible negative nature of the audit fee/NAS relationship as he

investigated “whether the audit is used as a loss leader and-or whether there are reductions

in audit costs which are returned to clients via lower audit fees” (Firth, 1997, p.52). A

negative relationship could however not be proven as the applied regression model displayed

a positive relationship (Firth, 1997). An argument for a negative relationship between audit

fees and NAS is the potential knowledge spillover (Krishnan & Yu, 2011) that can occur

when the audit and the NAS is conducted by the same firm. The notion is that information

gained from one service is transferred to the other service and thereby generating an

aggregated cost reduction. Krishnan and Yu (2011) finds a significant negative relationship

between audit fees and NAS and confirms the existence of knowledge spillover. In

accordance with recent research a negative relationship between audit fees and non-audit fees

is expected.

18

Audit fee premium

Big 4 audit firms can charge an audit fee premium due to the belief that Big 4 provide a

higher audit quality than non – Big 4 firms (Ebrahim, 2010). The studies conducted on the

audit fee premium area based on a single-country sample have garnered mixed results

regarding the existent and size of the audit fee premium (Choi et al., 2008). Although the

majority of studies have confirmed an audit fee premium for Big 4 firms (Caneghem, 2010;

Ebrahim, 2010) there are still studies that failed in doing so (Simunic, 1980) and studies with

mixed results depending on the regression model that was applied (Francis & Simon, 1987).

Ebrahim (2010) studied the effects of the implication of the Sarbanes-Oxley act (SOX) on the

audit fee premium and found a significant shift in the audit fee premium the initial years after

the SOX introduction which was later to be reduced in the third year after the introduction.

This shift can be attributed to the increased demand for audit services following an

implementation of new regulations; the need for more auditing services was more noticeable

for risky or non-profitable companies (Ebrahim, 2010)

Furthermore, in a cross-country sample study Choi et al. (2008) found that there is a fee

premium for Big 4 audit firms which decreases in countries with stronger legal liability

regimes and are more profound for small and medium sized firms than for large firms. Based

on the results of the majority of previous studies, an audit fee premium for the Big 4 audit

firms should exist on the Swedish audit market and therefore a positive relationship to audit

fees is expected.

Number of audit firms

In the Green Paper (European Commission, 2010), the European commission explores the

possibility of a mandatory joint audit of larger organizations to encourage the growth of small

and medium size audit firms. These joint audits would generate at least two signatures on the

audit report which implies that at least two audit firms would be sharing the responsibility for

the audit (Quick, 2012). Furthermore, Quick (2012) find support from several research papers

that audit quality and joint audits can increase competition and audit quality but might also

increase the monitoring cost and thereby the audit fee. However, in a study on the Danish

audit environment Thinggaard & Kiertzner (2008) concluded that joint audits with two

independent auditors sharing the work in a competitive environment might lower the audit

fees for large organizations.

19

André, Broye, Pong and Schatt (2011) found that France had higher expected audit fees than

the UK and attributed these finding to the legal requirements of joint audits and a six year

tenure required by the French organizations. Furthermore, in a study focused on the Swedish

audit environment Haapamäki, Järvinen, Niemi and Zerni (2011) found that the choice of two

auditors are associated with substantially higher fees then for companies who employ a single

auditor. In line with Quick (2012), André et al. (2011) and Haapamäki et al. (2011) a positive

relationship between joint audits and audit fees is expected.

Number of signing auditors

When the auditor signs off on an audit report the information issued by the client is

legitimized and the auditor does “transform the client's inherently untrustworthy explanation

into an inherently trustworthy one” (Pentland, 1993, p.613). The ritual of signing off on the

audit report can thereby be described as an act aimed to provide comfort and trust in the

financial statements issued by companies. The assurance provided by the act of signing off

can thereby be deemed essential from an agency theory perspective; the trust provided from

the auditor lacks substantive use for a stakeholder unless the auditor has in fact signed off on

the audit report.

Furthermore, Pentland (1993) describes the procedure of signing the audit report as a way of

putting your professional identity on the line, stating that by signing the audit report the

auditors puts their jobs on the line due to the risk of facing litigation charges. The existence of

an joint audit should thereby from an assurance perspective provide more comfort for the

stakeholder since an joint audit implies a shared responsibility (Quick, 2012) where more than

one auditor would run the risk of litigation. Hence, as the number of signing auditors increase

so should the time and effort spent on the audit which in turn would increase the charged audit

fee. Therefore, a positive relationship between the audit fees and the number of signing

auditors is expected.

20

2.4 Synopsis

Table 1

Synopsis of previous research and expected relationships

Factors affecting audit fees Est. Previous literature Accounting complexity

Proportion of financial assets - (Power, 1999)

Impairments of intangible assets + (Bens et al., 2011)

Accruals + (Srinidhi & Gul, 2007)

Provisions + (IAS 37; Bens & Johnston, 2009; Moehrle, 2002; Peek,

2004)

Free cash flow + (Nikkinen & Sahlström, 2004)

Inherent risk + (Hay et al., 2006; Stice, 1991; Taylor & Simon, 1999)

Business complexity

Media exposure (visibility) + (Brammer & Millington, 2005; Brammer & Pavelin,

2006; Roberts & Dowling, 2002)

Mergers and acquisitions + (Angwin, 2007; Cartwright & Schoenberg, 2006; Meglio

& Risberg, 2010)

Executive compensation + (Chng et al., 2012; Morse et al., 2011)

Size of the audited firm + (Menon & Williams, 2001; Simon & Francis, 1988;

Simunic, 1980)

Number of employees situated

abroad

+ (Menon & Williams, 2001; Simon & Francis, 1988;

Waresul Karim & Moizer, 1996)

Ownership structure - (Fleming et al., 2005; Jensen & Meckling, 1976;

Nikkinen & Sahlström, 2004)

Assurance

Non-audit fees - (Firth, 1997; Krishnan & Yu, 2011)

Audit fee premium + (Caneghem, 2010; Ebrahim, 2010; Francis & Simon,

1987)

Number of audit firms + (Quick, 2012; André et al., 2011; Haapamäki et al.,

2011)

Number of signing auditors + (Pentland, 1993; Quick, 2012)

Table 1 presents the variables, the estimated results for each variable and the previous research on which the estimations are based.

2.5 Hypotheses

Based on the previous research performed on the area of audit fee determinants, applying the

audit fee model initially created by Simunic (1980), this study aims to test the relationship

between problematic audit areas and audit fees. Agency theory states that as the agent and

principal might have diverging interests, a need for monitoring arises and as discussed earlier

in this chapter the monitoring need can be affected by different problematic factors. These

problematic monitoring factors can affect the basic assumption of audit fee determinants. The

basic assumption states that the audit fee differences can be twofold, either due to disparities

in the time spent on the audit or due to differences in the charged price (Simunic, 1980). The

three problematic monitoring groups will influence the audit fees through the increased need

for time spent on the audit assignment (e.g. due to the size of the company) or through a

higher audit pricing (e.g. due to need for expert knowledge). Hence, if this assumption holds

21

the problematic monitoring factors should affect the audit fee. We pose a generated

hypothesis based that relates to accounting complexity, business complexity and assurance:

Hypothesis 1: Audit fees increase related to accounting complexities, business complexities

and assurance.

The best interest of the agent and principal can easily diverge, this leads to an agency

problem. An agency problem increases the need for monitoring and the monitoring cost has

been found to be higher as the agent’s and principal’s interests significantly diverge (Fleming

et al., 2005). An extreme example would be a company where a majority owner control 100

% of the voting rights, according to theory on separation of ownership and control (Fleming et

al., 2005) the monitoring cost would be low in this example. The auditor has to assure the

principals that the agent is acting in their best interest and when the interest significantly

diverge it is reasonable to expect that the auditor has to devote more time to the audit

assignment thereby increasing the monitoring cost. The audit fee is an important part of the

monitoring cost, hence when the interest diverges due to a diversified ownership structure the

audit fee will be affected. Hence, we pose a hypothesis based on agency theory and ownership

structure differences:

Hypothesis 2: Audit fee determinants importance increases between different ownership

structures

22

3 RESEARCH METHOD

Chapter 3 presents the methodological design utilized for this study. The chapter begins with

a presentation of the overall research design (3.1) and the methodological considerations and

choices made in association with the overall research design. Following the overall research

design is a description of the data collection and the sample selection (3.2) and a description

of the variables measurements (3.3). Specifications of both the applied models are presented

in 3.4. The final section of the chapter presents the statistical method used in the study.

3.1 Overall research design

In order to test the relationship between the dependent variable and the independent variables

a quantitative approach is adopted as this approach is more suitable for handling and drawing

general conclusions from large quantities of data. The results generated will function as an

indicator of differences in audit fee determinants for ownership structures. As it is most suited

for quantitative research, a deductive approach is utilized. A deductive approach entails the

deduction of hypothesizes based on previous theoretical knowledge which is then tested

empirically (Bryman & Bell, 2011). Based on this approach two following hypothesizes have

been generated to solve the overall research aim. Data was collected for 17 variables from

annual reports from both primary and secondary sources. The data was collected for

companies listed on the Stockholm Stock Exchange from year 2001 to 2010, in total the

sample consisted of 2213 firm-years.

We apply a pooled analysis to combine all the collected data in to one single dataset. We then

tested the correlation between the dependent and independent variables through the

application of a bivariate analysis (Pearson correlations test). To test the hypotheses multiple

regression models were employed, the primary regression model tested the first hypothesis

and was then modified to test the second hypothesis. Thereafter robustness tests were

performed on both the primary and secondary regression model.

3.2 Sample selection and data sources

The sample consists of annual reports from the listed companies on the Nasdaq OMX

Stockholm Stock Exchange during the period 2001-2010. Publicly listed companies provides

an accessible base for data collection through the annual reports, which is public information,

while private companies annual reports can be harder to attain. Furthermore, the auditing of

public and private companies differs in the sense that different standards applies, in Sweden

23

as for all EU members publically listed companies are required to follow IFRS standards in

their financial reporting (Zerni, 2012). “Whereas Swedish private companies are also allowed

to follow IFRS standards in their consolidated financial statements, they tend to follow

national standards” (Zerni, 2012, p.318). The use of publicly listed companies could enhance

the generalizability of the study as the same standards apply for all EU members.

A timeframe of ten years is used due to the possible fluctuations in the audit fees that can

occur when the economy is in a recession or when new regulations are implemented. For

example, the introduction of IFRS in 2005 put upwards pressure on audit fees the years

subsequent the implementation (Kim et al., 2012). The timeframe stretches from 2001-2010

but not all companies were listed during the entire timeframe. However, excluding companies

that were not listed throughout the timeframe can lead to survival bias, which is avoided by

including these companies.

The total number of firm-years for the specific time period is 2671. However, financial

institutions, real estate companies and investment companies have been excluded from the

sample. Financial institutions have been excluded from the sample due to the unique

characteristics of the financial statements issued by financial institutions which makes them

less comparable to other industries. Real estate and investment companies are also excluded

from the sample due to the nature of their business which possibly could affect the studies

reliability negatively. The financial reports of investment- and real estate-companies tend to

have a different structure which makes a large number of independent variables untestable

(e.g. accruals and inherent risk). Furthermore, firms with total assets of less than 10 million

SEK were excluded to remove the effects that extremely small companies can have on the

results. After these exclusions the sample consisted of 2213 annual reports. In Table 2 the

sample description is specified.

Table 2

Sample description

Number of annual reports used in the total sample sorted by sector

Sector Population Sample % of population

1 Industrial 1242 1242 100 %

2 Trade 241 241 100 %

3 Services 733 730 99,6 %

4 Finance 455 0 0 %

TOTAL 2671 2213 82,9 %

Table 2 presents the number of annual reports in the total sample and how the sample has been transformed.

24

The data was collected from both primary and secondary sources. For the variable provision

data was collected directly from the annual reports whilst the other variables where collected

from a secondary database. The secondary database had collected the data manually from

DataStream, in instances where the necessary data had not been available in DataStream the

data was collected manually from annual reports. The reason for using secondary data to such

a large extent is twofold; to attain a larger sample and to more easily define what kind of data

was necessary for the study. Using secondary data can however be problematic, secondary

data can come with some limitations. Bryman & Bell (2011, p.320) lists some potential

limitations:

- Lack of familiarity with the data

- Complexity of the data

- Lack of control over data quality

- Absence of key variables

To limit these potential problems we devoted a period of time getting familiar with the data to

understand which information the different variables contained, how it was coded and

scanning for potential errors in the dataset. Regarding the lack of control over data quality it

should be noted that the data have been collected in research purposes not attributable to this

specific research area i.e. audit fees but to the accounting and finance area as a whole.

However, since the data have or will be utilized in other research projects it is reasonable to

assume that the data is of high quality. However, there is always a potential risk for errors

when collecting data from annual reports due to the human factor. The collection process did

however contain a low degree of interpretation which should limit the risk of error or diverse

observations.

The primary data regarding the variable provisions (PROV) was collected from annual

reports, when collecting the data we gathered information regarding:

- Opening balance for the year

- The change for the current year

- Closing balance for the year

- Cash flow for the year

- Reversals for the year

25

This information was gathered for total provisions, restructuring charges, warranties,

litigations, earn out and other provisions. We used a data collection protocol to ensure the

reliability of the data by limiting the risk of inter-observer inconsistencies which can

described as the heightened risk of different judgments when there is more than one observer

collecting the data (Bryman & Bell, 2011). The data collection was operationalized through

the use of multiple key words, i.e. provisions, restructuring, warranties, litigation, earn out,

legal and reserves. If none of these keywords yielded results the balance sheet, income

statement and notes was checked to find the necessary information.

The sample was then transformed using winsorization. Winsorization is a method for

modifying values and to treat the problems arising from extreme outliers, the extreme value is

replaced by a cut-off value to limit the skewness of the data (Gregoire & Schabenberger,

1999). We winsorized all the unbound variables which are endless at the 2nd

and the 98th

percentile thereby replacing all lower and higher values with the cut-off value.

3.3 Measures

3.3.1 Dependent variable

The dependent variable used in this study is audit fees. As with most previous research within

this is area the audit fees are measured as the natural logarithm of the total audit fees

(LOGAUDITFEE) (e.g. Gul et al., 1998; Taylor & Simon, 1999; Waresul Karim & Moizer,

1996) to achieve a better linear fit to the other variables in the regression model.

3.3.2 Independent variables

To test the audit fee determinants 16 independent variables are applied. The measurement of

the 16 independent variables is presented below. The independent variables and their

measurement are also presented in Table 3.

Accounting complexities

Monetary assets should be easy to verify, hence a company with a large share of financial

assets should require less audit effort. We measure the proportion of financial assets

(FINANCE) as the total sum of monetary assets divided by total assets at year end.

Impairment of intangible assets (IMPAIR) is measured as the total impairment of intangible

assets divided by total assets at the end of the year. Accruals (ACCRUAL) is measured as the

change in non-assets less the change in non-debt liabilities where non-cash assets is defined as

26

total assets less cash and short time investments and non-debt liabilities is defined as total

liabilities less debt as developed in a study by Dechow, Richardson & Sloan (2008).

Provisions (PROV) is measured as the annual change in provisions divided by total assets at

year end. Free cash flow (FCF) is measured as income less total accruals divided by total

assets; this measurement has been used in a study by Nikkinen & Sahlström (2004). Inherent

risk (IRISK) is associated with an increased risk of error in some parts of the financial

statement which will requires more audit effort, earlier research have predominately measured

the inherent risk in three different ways; Inventory divided by total assets, receivables divided

by total assets or inventory plus receivables divided by total assets (Hay et al., 2006). For a

more comprehensive measurement inherent risk will be measured as inventory plus

receivables divided by total assets.

Business complexities

A firm that is featured heavily in the media spotlight could be scrutinized more than a firm

which is not featured as often in media. To test this assumption, the number of times

mentioned in media during one year is used a measurement of total visibility (EXPOSURE).

EXPOSURE is measured as the frequency of times mentioned in the following printed media;

Affärsvärlden, Aftonbladet, Dagens Industri, Dagens Nyheter, Expressen, Göteborgsposten,

Privata Affärer, Svenska Dagbladet, Sydsvenska Dagbladet and Veckans Affärer. The

frequency of times mentioned have then been divided into decile groups and values from 1 to

10 are given where 10 means the highest exposure. Mergers and Acquisitions (M&A) is given

an indicator value; an indicator value of 1 is given if the acquisition of goodwill plus the

acquisition of other intangible assets divided by the total assets at year end exceeds ten per

cent, the acquisition value is measured as the difference between in- and out-going balance

after impairments and amortizations. If the value does not exceed ten per cent, a value of 0 is

assigned.

Stock options to CEO (EXCOMP) is also given an indicator value, 1 of the audited firm has

issued a stock option program to the CEO and 0 if not. Client size is the most common used

variable used in audit fee studies and in line with most previous research we measure client

size as the firm’s total assets (Hay et al., 2006). The variation between the smallest and largest

firms’ can be quite substantial so in order to achieve a better linear relationship with audit fees

the natural logarithm of total assets (LOGTOTASS) is used. The audit complexity is

previously commonly tested variable; earlier studies have tested the complexity through

27

different proxies, most used proxies have been number of subsidiaries or number of foreign

subsidiaries (Hay et al., 2006). A previously untested proxy for complexity is the number of

employees situated abroad; the assumption is that a large geographical spread will increase

the audit complexity. Hence, the number of employees situated abroad divided by the number

of total employees (FOROP) is used as a proxy for complexity. The ownership variable

(OWNERS) is measured as the percentage of voting rights owned by the largest shareholder.

Assurance

As with total assets the variation between the largest and smallest firms’ total non-audit fees

can be quite substantial and order to achieve a better linear relationship with audit fees the

natural logarithm of non-audit fees (LOGNONFEE) is utilized. Previous research (e.g. Choi et

al, 2008; Francis & Simon, 1987) indicates that Big 4 – auditors can charge an audit fee

premium (PREMIUM) which will increase the total audit fee. This is due to the implied

higher audit quality offered by a Big 4 – firm. Therefore, we use an indicator variable equal to

1 if a company uses a Big 4 audit firm and 0 if not. If a company employs more than one audit

firm (NUMFIRM) the total audit fee should increase. Hence, if a company employs more than

one audit firm an indicator value of 1 is given, if the company only employs one auditor a 0 is

given. Based on the same argument the variable which measures the number of signing

auditors (NUMAUD) is also given an indicator value of 1 if the audit report is signed by more

than one auditor and 0 otherwise. The two variables are mutually exclusive in the audit fee

model, i.e. if a firm-year value employs more than one audit firm and at the same has more

than one signing auditor it is only assigned a 1 for the NUMFIRM variable to avoid the

effects of causality. If a company is assigned a 1 for NUMAUD all auditors will be from the

same audit firm. Table 3 specifies how the variables are measured and the expected

relationship to audit fees.

28

Table 3

Presentation of the variables, estimations and variable measurement

Variables Name Est. Measurement

Dependent variable

Audit fee LOGAUDIT

FEE

Natural log of audit fees

Independent variables

Accounting complexity

Proportion of financial assets FINANCE - Sum of monetary assets divided by sum

of total assets at year end

Impairment of intangible assets IMPAIR + Sum of impairment of intangible assets

divided by total assets at year end

Accruals ACCRUAL + Change in non-cash asset less the

change in non-debt liabilities

Provisions PROV + Annual change in provisions divided by

total assets.

Free cash flow FCFA + Total sum of free cash flow for the

fiscal year divided by total assets

Inherent risk IRISK + Sum of inventories plus accounts

receivables divided by sum of total

assets at year end

Business complexity

Media exposure (visibility) EXPOSURE + Total times mentioned in media per

year, divided in to decile groups

Mergers and acquisitions M&A + (Indicator variable), (1) if company

participated in a M & A during the year

based on an increase in goodwill and

other intangible assets, (0) if not

Executive compensation EXCOMP + (Indicator variable), (1) if company

issued stock options to the CEO, (0) if

not

Size of the audited firm LOGTOT-

ASS

+ Natural log of total assets at year end

Complexity – Number of employees

situated abroad

FOROP + Employees situated abroad divided by

total employees

Ownership structure OWNERS - The percentage of voting rights owned

by the largest shareholder

Assurance

Non-audit fees LOGNON-

FEE

- Natural log of non-audit fees per year

Audit fee premium PREMIUM + (Indicator variable), (1) if auditor is a

Big 4 firm, (0) if not

Number of audit firms NUMFIRM + (Indicator variable), (1) If more than

one audit firm, (0) if not

Number of signing auditors NUMAUD + (Indicator variable), (1) If more than

one signing auditor, (0) if not

Table 3 presents the variables, the variables expected relationship to audit fees and the measurement of each individual variable.

29

3.4 Audit fee model specification

Based on the derived hypothesizes this study examine the audit fee determinants in different

ownership structures with aid of various independent variables. The model used in the study is

based on previous research (Francis & Wang, 2005; Nikkinen & Sahlström, 2004; Stanley,

2011) and modified to test audit fee determinants in different ownership structures. The

primary audit fee determinant model is presented below:

LOGFEEt = α +β1 FINANCEt + β2IMPAIRt + β3ACCRUALt + β4PROVt +

β5FCFAt + β6IRISKt + β7EXPOSUREt + β8M&At + β9EXCOMPt +

β10LOGTOTASSt + β11FOROPt + β12 OWNERSt + β13LOGNONFEEt +

β14PREMIUMt + β15 NUMFIRMt+ β16NUMAUDt

When testing whether the ownership structure affects the audit fee determinants the

OWNERS variable was removed as an individual independent variable and instead an

interaction variable labelled OWNCON. The secondary regression model adjusted for

ownership structure is presented below:

LOGFEEt = α +β1FINANCEt + β2IMPAIRt + β3ACCRUALt + β4PROVt +

β5FCFAt + β6IRISKt + β7EXPOSUREt + β8M&At + β9EXCOMPt +

β10LOGTOTASSt + β11FOROPt +β12LOGNONFEEt + β13PREMIUMt +

β14NUMFIRMt+ β15NUMAUDt β16FINANCEt * OWNCON +

β17IMPAIRt * OWNCON + β18ACCRUALt * OWNCON + β19PROVt *

OWNCON+ β20FCFAt * OWNCON + β21IRISKt * OWNCON +

β22EXPOSUREt * OWNCON + β23M&At * OWNCON+ β24EXCOMPt *

OWNCON + β25LOGTOTASSt * OWNCON + β26FOROPt *

OWNCON+ β27LOGNONFEEt * OWNCON + β28PREMIUMt

*OWNCON + β29NUMFIRMt * OWNCON +β30NUMAUDt * OWNCON

3.5 Statistical tests

The statistical testing is performed in IBM SPSS Statistics. To test for the existence of

correlation between the dependent and independent variables a bivariate analysis was

performed. Highly correlated independent variables can infer the effect of an individual

variable on the dependent variable i.e. multicollinearity; a bivariate test can help to predict the

existence of multicollinearity. To ensure that the generated results cannot be assigned to the

30

existence of highly correlated independent variables (multicollinearity) we performed a

Pearson correlations test. To avoid the effects of highly correlated independent variables we

also tried to detect the existence of multicollinearity through a Variance Inflation Factor (VIF)

test, if the observed value from the VIF-test is above 10 the independent variable should be

removed (Lind, Marchal & Wathen, 2010).

A multiple regression model was employed to test the connection between audit fees and the

audit fee determinants; a multiple regression model can be used to find a relationship between

a dependent variable and independent variables (Lind et al., 2010). Consistent with previous

research that have focused on audit fee determinants (e.g. Francis & Wang, 2005; Simunic,

1980; Stanley, 2011) we used a linear ordinary least squares regression model. The regression

model was then modified to test the second hypothesis regarding the audit fee determinants

for different ownership structures; the OWNERS variable was removed and replaced by an

interaction variable that took the value of 1 if the largest shareholder owns 50 % or more of

the voting rights and 0 otherwise.

Robustness tests were performed on both the primary and secondary regression model. We

extended the analysis of the primary regression model through testing of each audit fee

determinant group (Accounting complexity, Business complexity and Assurance) to examine

how much of the variation in audit fees can be explained by each individual group. The

primary regression model was then tested by excluding the independent variables that

displayed a high correlation to each other (LOGNONFEE, LOGTOTASS and EXPOSURE).

This test was performed to examine the individual variables effect on the result. The

secondary regression model was modified with a cut-off level at 35 % instead of 50 % to test

the effects on less concentrated ownership.

31

4 RESULTS

The following chapter presents the results of the empirical tests introduced in the previous

chapter. Firstly the descriptive statistics (4.1) for the variables in the audit fee models are

displayed followed by the correlations between the dependent and independent variables and

a discussion on the interaction between the independent variables (4.2). Section 4.3 presents

and analyses the results from the primary and secondary regression model based on the

theoretical framework. Section 4.4 concludes the chapter with a presentation and analysis of

the results from the robustness tests of the primary and secondary regression models.

4.1 Descriptive statistics

Table 4 presents the descriptive statistics for the variables included in the audit fee model.

The natural logarithm of audit fees had a mean of 7,5 with a standard deviation of 1,43.

Executive compensation has a mean of 0,42 with a standard deviation of 0,49, the range is

between 0 and 1 as it is an indicator variable.

Table 4

Descriptive statistics

N Mean Median Std. Deviation Minimum Maximum

LOGFEE 2213 7,50 7,33 1,43 3,00 11,00

FINANCE 2213 0,17 0,11 0,17 0,01 0,74

IMPAIR 2213 0,01 0,00 0,04 0,00 0,21

ACCRUAL 2213 -1,22 0,03 19,72 -647,07 1,33

PROV 2213 0,03 0,00 1,69 -34,77 39,97

FCFA 2213 -0,21 0,00 0,81 -4,50 0,63

IRISK 2213 0,30 0,29 0,17 0,00 0,67

EXPOSURE 2213 5,43 5,00 2,82 1,00 10,00

M&A 2213 0,13 0,00 0,33 0,00 1,00

EXCOMP 2213 0,42 0,00 0,49 0,00 1,00

LOGTOTASS 2213 7,09 6,78 1,93 3,70 11,45

FOROP 2213 0,48 0,48 0,35 0,00 1,00

OWNERS 2213 0,32 0,27 0,20 0,00 1,00

LOGNONFEE 2213 6,71 6,62 1,60 2,00 10,00

PREMIUM 2213 0,91 1,00 0,28 0,00 1,00

NUMFIRM 2213 0,07 0,00 0,25 0,00 1,00

NUMAUD 2213 0,30 0,00 0,46 0,00 1,00

Table 4 presents the descriptive statistics for the dependent variable and the independent variables. The descriptive statistics presented is

Mean, Median, Std. Deviation, Minimum and Maximum. See Table 3 for measurement definitions.

For the indicators variables the standard deviation is larger than the mean except for the

PREMIUM variable which shows a large degree of firms having a Big 4 auditor (0,91).

32

4.2 Correlations

The correlations between the variables utilized in the audit fee models are presented in Table

5. The correlation matrix displays the different individual correlations between two variables.

The dependent variable is LOGFEE and the independent variables are FINANCE, IMPAIR,

ACCRUAL, PROV, FCFA, IRISK, EXPOSURE, M&A, EXCOMP, LOGTOTASS, FOROP,

OWNERS, LOGNONFEE, PREMIUM, NUMFIRM and NUMAUD.

The strongest correlations in relation to LOGFEE are displayed by the variables

LOGNONFEE, LOGTOTASS and EXPOSURE. LOGFEE and LOGNONFEE have a

positive correlation of 0,84 indicating a strong correlation between the two variables. Between

LOGFEE and LOGTOTASS the positive correlation is 0,91 which indicate a very strong

relationship and the correlation is the strongest between any two variables in the matrix.

LOGFEE and EXPOSURE have a slightly less strong correlation than the other two variables,

the positive correlation between LOGFEE and EXPOSURE is 0,65 which indicate a

moderately strong correlation.

Table 5

Correlations amongst variables in the audit fee model. Part 1(2).

1 2 3 4 5 6 7 8 9

1 LOGFEE 1,000

2 FINANCE -0,420 1,000

3 IMPAIR -0,077 0,070 1,000

4 ACCRUAL 0,063 -0,034 -0,019 1,000

5 PROV -0,002 0,021 0,057 -0,005 1,000

6 FCFA 0,229 -0,165 -0,275 0,334 -0,019 1,000

7 IRISK 0,101 -0,358 -0,048 0,035 0,005 0,030 1,000

8 EXPOSURE 0,653 -0,126 -0,032 0,063 -0,016 0,120 -0,121 1,000

9 M&A 0,014 -0,103 0,088 -0,022 0,060 0,041 -0,072 -0,009 1,000

10 EXCOMP 0,096 0,040 0,015 0,012 0,004 -0,013 -0,027 0,065 0,024

11 LOGTOTASS 0,914 -0,380 -,150 0,083 -0,002 0,273 -0,004 0,698 -0,025

12 FOROP 0,435 -0,095 0,003 0,046 0,009 0,067 -0,093 0,311 0,040

13 OWNERS 0,109 -0,104 -0,074 0,001 -0,013 0,070 0,165 -0,075 -0,049

14 LOGNONFEE 0,838 -0,329 -0,045 0,039 0,007 0,148 -0,007 0,634 0,021

15 PREMIUM 0,142 -0,107 -0,032 0,031 0,003 0,119 -0,040 0,098 0,031

16 NUMFIRM 0,114 -0,055 0,028 0,009 -0,006 -0,011 -0,021 0,067 -0,028

17 NUMAUD -0,031 -0,059 -0,065 -0,021 -0,011 -0,017 0,180 -0,020 -0,033

33

Correlations amongst variables in the audit fee model. Part 2(2).

10 11 12 13 14 15 16 17

10 EXCOMP 1,000

11 LOGTOTASS 0,064 1,000

12 FOROP 0,005 0,387 1,000

13 OWNERS -0,078 0,079 -0,067 1,000

14 LOGNONFEE 0,108 0,828 0,396 0,024 1,000

15 PREMIUM 0,051 0,168 0,092 -0,052 0,141 1,000

16 NUMFIRM 0,012 0,131 -0,033 0,099 0,119 -0,226 1,000

17 NUMAUD 0,040 -0,027 -0,072 0,035 -0,026 0,071 -0,176 1,000 Table 5 present the Pearson correlations for the 17 variables. The sample covers the years between and 2001-2010 and consists of 37 621

firm-year observations. Bolded correlation is significant at the 0.05 level (2-tailed).

A number of variables have a negative correlation with LOGFEE, suggesting that if LOGFEE

increases the other pairwise variable will decrease in value and vice versa. The correlation

between LOGFEE and FINANCE is negative, having a value of -0,42 which indicates a

moderately strong negative relationship. The other variables that experiences a negative

correlation in relation to LOGFEE is IMPAIR, PROV and NUMAUD.

4.2.1 Interaction between the independent variables

The bivariate analysis displayed indications of strong correlation between LOGTOTASS,

LOGNONFEE and EXPOSURE. It is important to consider the underlying reasons for why

these correlations exist and how it affects the results of the audit fee determinant model.

LOGTOTASS had a correlation of 0,83 to LOGNONFEE which indicates a strong correlation

between the two variables. At the same time both independent variables showed a strong

correlation to LOGFEE. Considering the findings of previous research on audit fee

determinants (e.g. Menon & Williams, 2001; Simon & Francis, 1988; Simunic, 1980) which

suggests that the size of the audited company has a large impact on audit fees, it would be

plausible to expect that the relationship between LOGTOTASS and LOGNONFEE display

similar characteristics. Since, LOGFEE displays a strong correlation to both LOGTOTASS

and LOGNONFEE it seems reasonable to expect that as the company grows in size so does

both the audit fee and the non-audit fee. Hence, LOGTOTASS and LOGNONFEE should

display a strong correlation since a larger company should both have the need as well as the

resources to employ non-audit fee services.

As mentioned above, EXPOSURE also displays a strong correlation to LOGTOTASS and

LOGNONFEE. Theoretically, a larger company should be more interesting from a public

perspective as the number of stakeholders is larger and the possible ramifications in case of a

company failure can be quite substantial from a society perspective. Thus, a large company

34

should be subjected to more media attention which would explain the strong correlation

between EXPOSURE and LOGTOTASS. The strong correlation between EXPOSURE and

LOGNONFEE can be explained through the prism of auditee size by expanding on the

previous discussion on LOGTOTASS and LOGNONFEE which suggested that large

companies require more non-audit fee services. By inverting that relationship it is reasonable

to suggest that a company with high non-audit fees is in fact a large company and thereby in

turn should receive more media attention (as the correlation between LOGTOTASS and

EXPOSURE is strong) which could explain the moderately strong correlation between

EXPOSURE and LOGNONFEE.

FCFA and ACCRUAL display a weak relationship (0,33) which could indicate a problem

with multicollinearity. The probable explanation for this correlation could be found in the

applied measurement definition. Free cash flow is defined as income less accruals as

developed by Dechow et al. (2008). However, this means that there is a direct relationship

between FCFA and ACCRUAL that could explain the correlation between the two variables.

The use of accrual accounting will also affect the available free cash flow at year end due to

the shift in actual funds available and thus strengthening the correlation between the two

variables. The different correlations (except the correlation between FCFA and ACCRUAL)

with any degree of significance can in all cases be attributed to the size of the company,

indicating that the size of the auditee has a large impact, both directly and indirectly, on audit

fees.

IRISK and FINANCE have negative correlation of -0,36 indicating that the proportion of

financial assets has a weak negative relationship to the IRISK variable. Most pairwise

correlations display either a low or non-meaningful correlation. Therefore multicollinearity

does not appear to be an issue for most independent variables, i.e. one variable is not totally

predicted depending on another variable. Strong correlations can indicate a problem with

multicollinearity; to ensure that the results were not affected by the existence of

multicollinearity a VIF-test was performed. The VIF-test showed no indication of

multicollinearity; hence there was no need to exclude any of the independent variables4.

4.3 Regressions

The results from the multiple regression models are presented and analysed below.

4 The result from the VIF-test is presented in Appendix 1

35

4.3.1 Primary audit fee model

Table 6 presents the result from the primary audit fee regression model; the results show that

the primary audit fee model is highly significant. The observed p-value is 0,000 which

indicates a very strong significance. The closer the R-squared value is to 1, the more variation

in the dependent variable is explained by the independent variables. The adjusted R-squared

value is 0,876 which indicate that the variation in the dependent variable LOGFEE is highly

explained by the variation in the independent variables. An R-squared value of 0,876

indicates that the applied audit fee model explains approximately 88 % of the variance in

audit fees. The unexplained variance can be twofold, either from untested independent

variables or from circumstances that the model fails to explain. The primary regression model

indicates some discrepancies between the expected relationships and the actual observed

results. If the P-value is less than 0,05 the independent variable indicates a significant relation

to LOGFEE.

Table 6

Primary regression model and bivariate test for audit fee determinants

Independent variables Est. Beta t Sig. Bivariate test Adj. R2

(Constant)

26,12 0

FINANCE - -0,023 -2,408 0,016 -0,420 17,6%

IMPAIR + 0,047 5,695 0,000 -0,077 0,5%

ACCRUAL + -0,015 -1,851 0,064 0,063 0,4%

PROV + -0,007 -0,918 0,359 -0,002 0,0%

FCFA + 0,009 1,053 0,292 0,229 5,2%

IRISK + 0,113 12,732 0,000 0,101 1,0%

EXPOSURE + 0,014 1,237 0,216 0,653 42,6%

M&A + 0,020 2,519 0,012 0,014 0,0%

EXCOMP + 0,020 2,523 0,012 0,096 0,9%

LOGTOTASS + 0,684 38,908 0,000 0,914 83,5%

FOROP + 0,083 9,584 0,000 0,435 18,9%

OWNERS - 0,014 1,696 0,090 0,109 1,1%

LOGNONFEE - 0,223 15,461 0,000 0,838 70,1%

PREMIUM + -0,006 -0,757 0,449 0,142 2,0%

NUMFIRM + -0,009 -1,041 0,298 0,114 1,3%

NUMAUD + -0,020 -2,463 0,014 -0,031 0,0%

Adjusted R2

87,7%

Model F (p-value) 0,000

Table 6 reports the results of the primary regression model and the bivariate test explaining audit fees. The sample comprises 37 621 firm

year observations and covers the time period 2001-2010. The variables with a value under 0,05 in the Sig. column are significant. For the full

results from SPSS regression see Appendix 2.

36

The independent variables that display a positive significant relationship to LOGFEE are

LOGNONFEE, M & A, EXCOMP, LOGTOTASS, FOROP, IRISK and IMPAIR. The

variables that display the strongest significant relationship to LOGFEE are LOGTOTASS

LOGNONFEE and IRISK. For LOGTOTASS this indicates that as the company grows in

size, measured by total assets, the audit fee will increase.

FINANCE and NUMAUD display a significant negative relationship with LOGFEE. This

indicates that if the number of auditors signing the audit report is higher than one the audit

fees will be lower than for companies with just one signing auditor. The remaining variables

PREMIUM, EXPOSURE, PROV, FCFA, OWNERS, ACCRUAL and NUMFIRM display a

mix of negative and positive relations to LOGFEE that are not significant.

Accounting complexities

FINANCE (P=0,016) displayed a significant negative relationship to audit fees indicating that

as the proportion of financial assets to total assets increase the audit fees will decrease. The

expectation was a negative relationship, which is line with the findings, due to the assumption

that financial assets are easy verifiable. The valuation of financial assets seldom involves

subjective estimations which should indicate that the verification process is rather

straightforward i.e. not time-demanding.

According to IAS 36 (IASB, 2004) intangible assets requires annual impairment testing, the

impairment testing entails subjective judgment and according to Bens et al. (2011) the

standard leads to unverifiable accounts with low reliability. An account with low reliability

heightens the risk of material misstatements which from an agency theory perspective could

lead to information asymmetry and therefore an increased demand on monitoring (Nyberg et

al., 2010). Hence, the auditor either has to devote more time or specialist knowledge on

auditing impairments of intangible assets. The findings indicate support for the previous

discussion as the variable IMPAIR (P=0,000) displays a significant positive relationship to

audit fees which also was expected.

Accruals is an area where the interests of the agent and principal easily can diverge, the use of

accruals can lead to earnings manipulation, either intentionally or unintentionally, from

managers. As the interest of the agent and principals diverge the monitoring cost will increase

to make certain that the agent acts in the way that benefits the principal (Mitnick, 1973).

When the accruals increase there is a greater risk for the interests of agents and principals to

37

diverge which should therefore increase the monitoring cost. However, our findings suggest

that more accruals do not lead to an increase in audit fees as the relationship between audit

fees and ACCRUAL (P=0,064) is negative but non-significant. The result indicates that the

accruals already are of high quality without the need of auditor involvement which is in line

with the findings by Srinidhi & Gul (2007). The results from the bivariate analysis do display

a positive relationship between accruals and audit fees which suggest that the effect of

accruals is mitigated by the other variables in the primary regression model.

The result for PROV (P=0,359) indicates a non-significant negative relationship to audit fees.

The results provided no support for the estimation that provisions are positively related to

audit fees due to the subjective nature of provisions. The findings could indicate that earnings

smoothing is not regarded as a problem even though it might lead to information asymmetry.

This contradicts agency theory that proposes that the information asymmetry is problematic

and increases the monitoring cost. The bivariate analysis also suggested that provisions are

not related to audit fees thereby supporting the results from the primary regression model.

Free cash flow is expected to increase the conflicts between agents and principals when the

amount of free cash flow is substantial. Therefore, when the amount of free cash flow

increases the interest of the agent and principal are more likely to diverge and the monitoring

cost will increase. Thus, it is reasonable to expect a positive relationship between free cash

flow and audit fees. Although our findings do suggest a positive relationship for FCFA

(P=0,292) to audit fees it is not significant and the expectation cannot be supported. Our

findings support the results of previous research by Nikkinen & Sahlström (2004) whom also

found a positive non-significant relationship between FCFA and audit fees in the Swedish

setting. However, as Nikkinen & Sahlström (2004) results are based on a much smaller

sample our findings do offer increased support to the existence of a non-significant

relationship between free cash flow and audit fees.

In line with previous research (e.g. Hay et al., 2006; Stice, 1991) the findings indicate a strong

positive relationship between the inherent risk of a company and audit fees with the IRISK

variable having a P-value of 0,000. This suggests that the risk of material misstatements can

lead to an increase in the time spent on the audit assignment as well as employment of more

specialist knowledge. Material misstatements in the annual report can lead to information

asymmetry between the agent and principal. An issued annual report containing material

misstatements will not give the principal a true and fair view of the audited company. Hence,

38

the principal will require more monitoring to safeguard against material misstatements. From

an auditor perspective the risk of material misstatements can heighten the risk of litigation

(Taylor & Simon, 1999); it is therefore in the best interest of the auditor to delegate more time

and effort to the accounts with a high inherent risk.

Business complexities

The variable EXPOSURE (P=0,216) indicates a positive but non-significant relationship to

audit fees meaning that as the media visibility increases the effect on audit fees is uncertain.

Working under the assumption that media coverage will influence corporate reputation, which

is line with findings from Brammer & Millington (2005), these results are quite surprising. A

company which is more heavily featured in media will generally be more subjected to

scrutiny, potentially heightening the risk of litigation for the auditor, which should increase

the audit fee through an extended audit effort. However, (Brammer & Pavelin, 2006) states

that media exposure can increase the amount of available information for shareholders. As the

information available for shareholders increase the information asymmetry decreases. From

an agency theory viewpoint diminishing information asymmetry limits the need to monitor

the agent, thus lowering the monitoring cost. The findings might suggest that a trade-off

between the increased litigation risk and information asymmetry occurs, leaving the audit fee

unaffected.

An M&A is a complex event (Cartwright & Schoenberg, 2006) that has been exploited in

numerous manipulation scandals from management (Schilt & Perler, 2010). The principal, i.e.

the shareholders, wants to ensure that management does not undertake M&A’s with

inconclusive motives (Nguyen et al., 2012) which could be done by application of more

monitoring. The results from the primary regression model, regarding the M&A (P= 0,012)

variable, asserts a significant positive relationship between M & A and audit fees indicating

that the prior discussion is reasonable.

Executive compensation in the form of stock options to CEO is seen as an incentive that

promotes less aggressive behaviour compared to other incentive programs. Stock option

programs can be used as a mechanism to align the interest of the agent and principal.

However, according to Morse et al. (2011) up to 30 % of the incentive pay can be explained

by measurement rigging. Through the agency theory viewpoint, measurement manipulation

will diverge from the best interest of the principal resulting in a need for monitoring of the

39

executive compensation. Our findings displays a significant positive relationship between

EXCOMP (P= 0,012) and audit fees indicating that the existence of executive compensation

options does increase the audit fees. A possible explanation for this relationship can be that

the audit effort increases due to the risk of incentive program manipulation.

As previous research (e.g. Simon & Francis, 1988; Simon, 1985; Simunic, 1980; Waresul

Karim & Moizer, 1996) our findings conclude a strong significant positive relationship

between LOGTOTASS (P =0,000), and audit fees. As the company grows the audit fee

increases due to the demand for more time and effort spent on the audit engagement. The

auditor has to review the auditees accounts (Nikkinen & Sahlström, 2004) and as the size of

the audited company increases so does the complexity and the quantity of the financial

accounts. The bivariate test also supports the results from the primary regression model, the

bivariate test indicates that the size variable alone explains 83,5% of the variation in audit

fees.

The audit complexity increases as the audited company has a more geographical spread

(Menon & Williams, 2001) and to warrant a sufficient quality more time has to be assigned to

the audit object (Simunic, 1980). Previous research on complexity identifies a positive

relationship to audit fees (Simon & Francis, 1988; Taylor & Simon, 1999). Auditees’

operating in wider geographical spread through many different subsidiaries requires the

involvement of more audit teams. The audit team in the country where the parent company is

situated will require help from audit teams in the proximity of the various subsidiaries.

Thereby, the complexity of accumulating information for a sufficient audit increases and in

turn so does the monitoring cost. The geographical spread could also lead to information

asymmetry as it will be more difficult for an individual principal to gather information about

the company. Hence, the need for monitoring of the agent increases. Our findings indicate a

strong significant positive relationship between the variable FOROP (P= 0,000) and audit fees

suggesting that a geographical spread increases the audit complexity and the need for

monitoring.

In contrast to our expectations, OWNERS (P= 0,090) display a non-significant positive

relationship to audit fees indicating that as the concentration of voting rights increases so does

the audit fees. As one single shareholder owns a large portion of the company the risk of

divergence between agent and principals best interest should be limited as the majority owner

either is directly involved in the company or can choose employees who are trusted to limit

40

the need for monitoring. The results could therefore indicate that as one single shareholder

owns a large part of the company the possible impact in case of company failure would be

severe, i.e. the owner has a lot at stake in the company meaning that need for control and

auditing is greater.

Assurance

LOGNONFEE (P=0,000) displayed a strong positive relationship to audit fees meaning that

as the non-audit fee increases so does the audit fee which is not in line with the expected

relationship. Krishnan and Yu (2011) found to a negative relationship between audit fees and

non-audit fees due to the knowledge spillover meaning that an audit firm who also is hired to

perform NAS could use the information gathered through NAS in their audit work thereby

reducing the audit fee. However, the findings suggest that audit firms in the Swedish setting

separate their audit- and non-audit-services which leads to a situation where the knowledge

spillover is lost. The results are also more in line with earlier research performed on settings

more similar to the Swedish one, e.g. Firth, 1997. It is also reasonable to assume that as an

auditee receive more NAS the audit complexity and need for assurance increases thereby

affecting the audit fees positively. The lack of knowledge spillover could also indicate that the

auditee employs different companies for their audit services and for their NAS meaning that

knowledge spillover is not possible, this is however not the case for the applied sample. In the

sample the audit firm does take a large portion of the non-audit fees which raises question

about the validity of the knowledge spillover argument, at least for the Swedish setting. The

results from the bivariate test do also support the findings from the primary regression model.

The bivariate test indicated a strong relationship between audit fees and NAS.

The variable PREMIUM (P= 0,449) displays a negative, but non-significant, relationship to

audit fees suggesting that the employment of a Big 4-firm does not affect the audit fee which

is in contrast to what was expected. However, as previous research (Caneghem, 2010; Francis

& Simon, 1987; Simunic, 1980) has garnered mixed results this is not surprising. The results

from the bivariate analysis indicates that a positive significant relationship between audit fee

premium and audit fees do exist, hence the results from the primary regression model can

perhaps be attributed to the correlation with other variables.

Previous research finds that joint audits can increase competition and thereby audit quality but

also in extension increase the audit fee (Quick, 2012). Haapamäki et al. (2011) found results

41

indicating that in the Swedish setting the employment of two audit firms are associated with

higher audit fees than for companies employing just one audit firm. Based on this a positive

relationship was expected. However, our findings indicate a non-significant negative

relationship between NUMFIRM (P= 0,298) and audit fees. It is reasonable to expect that

competition between the appointed firms in a joint audit will lower the audit fee. However,

since the appointment of more than one audit firm should increase the audit quality and

thereby the assurance the audit fee will increase leading to an offset of the combined effect.

NUMAUD (P= 0,014) displays a significant negative relationship between the number of

signing auditors and audit fees. For assurance purposes a company can require the signing of

more than one auditor which would increase the monitoring cost. Nevertheless, when an audit

report is signed by more than auditor the litigation risk should decrease as the level of

assurance is higher. The result differs from the estimated positive relationship between the

number of signing auditors and the audit fee. The explanation for the difference can be

twofold; the signing of more than one auditor from the same firm does not increase the audit

effort and will not increase the cost of the audit, or the value of increased assurance that the

auditee gains from more than one signing auditor is less than the value of a reduced litigation

risk that the auditors gain from more than one signature.

Since seven of the 16 variables displayed strong positive significance to audit fees the first

hypothesis is accepted.

4.3.2 Secondary audit fee model

Table 7 present the results from the secondary regression model, where the primary audit fee

model is adjusted for ownership structure. The results display the difference between

ownership structures, for example by adding the betas for PREMIUM and PREMIUM *

OWNCON together the PREMIUM beta for companies where the largest shareholder owns

50% or more of the voting rights can be calculated. Hence, the PREMIUM beta for companies

where the largest shareholder owns at least 50 % of the voting rights is -0,003 + (-0,004) = -

0,007. The observed p-value is 0,000 which indicates a very strong significance. The adjusted

R-squared value is 0,877 meaning that the independent variables in the secondary regression

model explain 0,001 more of the variation in the dependent variable than the primary

regression model. If the P-value is less than 0,05 the independent variable has a significant

relationship to LOGFEE.

42

The secondary regression model adjusted for ownership structure indicated significant

difference for two of tested variables when the largest shareholder own at least 50 % of the

voting rights; EXCOMP * OWNCON and FOROP * OWNCON.

Table 7

Secondary regression model for audit fee determinants

Adjusted for ownership structure

Secondary audit fee model adjusted

for ownership structure

(Cut-off at >= 50%)

Change in Beta when ownership

concentration is at least 50%

(Independent variable*OWNCON)

Independent

variables

Standardized

coefficient

Beta t Sig.

Standardized

coefficient

Beta t Sig.

(Constant)

26,462 0,000

26,462 0,000

FINANCE -0,028 -2,683 0,007 0,014 1,267 0,205

IMPAIR 0,043 4,877 0,000 0,013 1,426 0,154

ACCRUAL -0,016 -1,854 0,064 0,008 0,895 0,371

PROV -0,007 -0,839 0,402 0,000 0,025 0,980

FCFA 0,012 1,238 0,216 -0,007 -0,681 0,496

IRISK 0,112 11,556 0,000 0,016 0,857 0,392

EXPOSURE 0,021 1,597 0,111 -0,034 -1,424 0,155

M&A 0,017 2,032 0,042 0,005 0,594 0,552

EXCOMP 0,011 1,299 0,194 0,024 2,188 0,029

LOGTOTASS 0,682 35,914 0,000 0,062 0,886 0,376

FOROP 0,068 7,146 0,000 0,056 3,667 0,000

LOGNONFEE 0,230 14,570 0,000 -0,108 -1,728 0,084

PREMIUM -0,003 -0,333 0,739 -0,004 -0,155 0,877

NUMFIRM -0,008 -0,809 0,419 0,001 0,057 0,954

NUMAUD -0,015 -1,670 0,095 -0,017 -1,461 0,144

Adjusted R2

87,8%

Model F (p-value) 0,000

Table 7 reports the results of the secondary regression model explaining audit fees adjusted for ownership structure. The sample comprises

37 621 firm year observations and covers the time period 2001-2010. The variables with a value under 0,05 in the Sig. column are

significant. . For the full results from SPSS regression see Appendix 3.

The remaining interaction variables displayed a non-significant difference for ownership

structure suggesting that the ownership structure does not have a large effect on the audit fee

determinants.

Executive compensation has a higher explanatory value for audit fees when the company has

one majority owner with 50 % or more of the voting rights. This is highlighted by the

significant positive relationship for the variable EXCOMP * OWNCON (P= 0,029). This is a

quite surprising relationship, when the largest shareholder owns at least 50 % of the voting

rights that individual have a large influence in appointing the board who in turn appoints the

43

CEO. Hence, the largest shareholder should appoint a trustworthy board who then appoints a

CEO they can trust i.e. the need for monitoring cost in regard to executive compensation

should decrease as the largest shareholder would trust the executive not to abuse the

compensation program. The reason for this relationship is however quite difficult to decipher

and might require further investigations to unravel the underlying explanation.

The FOROP * OWNCON (P= 0,000) variable displays a significant positive relationship

indicating that as the ownership is more concentrated the effect of foreign operations on audit

fees is accentuated. As the previous discussion on foreign operations above suggested, the

geographical spread can make it more difficult for an individual owner to gather information

about the company. When one shareholder owns a large portion of the company that

shareholder has a lot at stake in the company suggesting that the risk of information

asymmetry can be damaging. Hence, to minimize the information gap the large shareholder

would demand more monitoring thus increasing the audit fee. The geographical spread can

also lead to material misstatements as the complexity increases, as the risk for material

misstatements increase the auditor would want to devote more effort to avoid litigation risk. It

is reasonable to expect that an individual who owns a large portion of a company would be

more inclined to examine the auditor’s work than a small owner which would increase the

chance of detecting auditor mistakes. Thereby, the auditor should be more thorough to

minimize the litigation risk.

Noteworthy is that the variables that differ depending on ownership structure are related to

business complexities indicating that when a controlling owner owns 50 % or more of the

voting rights the auditor does devote more time to the business related complexities. Fan &

Wong (2005) suggested that an agency problem arises as one owner controls a large portion

of the voting rights, to alleviate the effect of the agency problem the controlling owner can

employ monitoring mechanisms to assure the other shareholders that their interests are

guaranteed. A concentrated ownership creates a situation where the controlling owner can

satisfy his or her own best interest (Fan & Wong, 2005). The results for the ownership

structure test suggests minor differences in auditor effort regardless of ownership structure

which could indicate that the auditor does not adjust the work effort despite the risk for

opportunistic behaviour. The increase in audit fees in relation to executive compensation

could, however, relate to various scandals where opportunistic behaviour has enabled

44

managers to receive large bonus from executive compensation programs requiring the auditor

to devote more time to that specific area.

The second hypothesis was rejected based on the observed results as the ownership structure

does not have a large effect on the audit fee determinants.

4.4 Robustness tests

Test of primary regression model per audit fee determinants group

The results from the primary regression model per audit fee determinants group is presented

in Table 8. The observed adjusted R-squared value for accounting complexity, business

complexity and assurance is respectively: Accounting complexity (R2 =0,207), Business

complexity (R2=0,844) and Assurance (R

2=0,703).

Table 8

Primary regression model modified and tested per audit fee determinants group

Independent variables Standardized coefficient

Beta t Sig.

Adjusted

R2

Model F

(p-value)

Accounting Complexities

(Constant)

110,376 0,000

FINANCE -0,414 -19,608 0,000

IMPAIR -0,013 -0,634 0,526

ACCRUAL -0,005 -0,226 0,821

PROV 0,011 0,543 0,587

FCFA 0,165 7,590 0,000

IRISK -0,055 -2,614 0,009 20,7% 0,000

Business complexities

(Constant)

51,580 0,000

EXPOSURE 0,013 1,025 0,305

M&A 0,028 3,195 0,001

EXCOMP 0,026 3,007 0,003

LOGTOTASS 0,866 66,941 0,000

FOROP 0,091 9,711 0,000

OWNERS 0,016 1,839 0,066 84,4% 0,000

Assurance

(Constant)

26,974 0,000

LOGNONFEE 0,829 67,301 0,000

PREMIUM 0,038 2,988 0,003

NUMFIRM 0,029 2,298 0,022

NUMAUD -0,010 -0,801 0,423 70,3% 0,000 Table 8 presents the results from the robustness test of the primary regression model.

The primary regression model was tested for the three audit fee determinants groups (Accounting complexity, Business complexity and

Assurance)

45

This indicates that business complexities and assurance explains a greater portion of the

variation in the dependent variable audit fees than accounting complexities does. One possible

explanation for the high R-squared value for business complexity is the size variable which

has a very high t-value. Size is also correlated to some extent with the other variables making

the effects of the size variable even greater. The risk for opportunistic behavior is quite

substantial for executive compensation and M&A’s meaning that the best interest of the agent

and principal might diverge hence increasing the monitoring cost. The R-squared value for

assurance is highly explained by the LOGNONFEE variable.

Assurance issues can be used to assure the principal that the agent is acting in his or her best

interest. As an auditor is required to provide a high level of assurance to the shareholders

(European commission, 2010) the audit effort devoted to assurance issues should be high.

Hence, the audit fee should have a strong relationship to assurance.

The R-squared value for accounting complexities displays a low explanatory power of the

variation in the audit fees. Auditors might have standardized procedures and protocols for

checking the financial accounts thereby decreasing the audit effort. However, the financial

accounts can be quite troublesome. It is within the financial accounts that the material

misstatements occur which can increase the information asymmetry between the agent and

principal which should increase the audit effort regarding accounting complexities. It is

important to note that a large part of the explanatory power is found within a few variables

(i.e. LOGTOASS and LOGNONFEE).

Exclusion of audit fee determinants from primary regression model

As the bivariate analysis showed a strong correlation between the variables LOGTOTASS,

LOGNONFEE and EXPOSURE a robustness test was performed. In the robustness test these

variables were individually removed to control if the result were affected. By excluding the

three independent variables the effects on the variation in the dependent variable can be

identified.

The results from the robustness test of the primary regression model5 showed that the

exclusion of all three variables reduced the R-squared value to 0,386 indicating that three

variables jointly have a large effect on the variation in the dependent variable. The exclusion

of one variable individually showed that by excluding LOGTOTASS the R-squared value was

5 The results from the modified primary regression models are presented in Appendix 4

46

reduced to 0,786, for LOGNONFEE it was reduced to 0,863 and for exposure the exclusion

reduced the R-squared value to 0,855. As the results for the exclusion of three variables

jointly showed such a large effect this could indicate that LOGNONFEE and EXPOSURE

contains elements affected by the size variable.

Test of change in ownership cut-off level

The second regression model was tested by changing the OWNCON variable cut-off from

50% to 35 %6. A controlling owner with at least 35 % of the voting rights can control a

company due to a large influence in the appointment of the board of directors and other voting

questions. Testing the secondary regression model for a 35 % cut-off level could provide

additional support for the findings that audit fee determinants can differ depending on

ownership structures. The modified secondary regression model displayed significant

relationships for several more variables than the secondary regression model at a 50% cut-off

level did. This suggests that the differences between ownership structures are more

accentuated as the largest shareholder does not have a majority of the voting rights. From the

viewpoint of agency theory this seems reasonable as the interests of the agent and principals

are more likely to diverge implying that the need for monitoring should increase.

The modified secondary regression model displays a significant positive relationship to

ownership structure for LOGTOTASS, FOROP and FINANCE indicating that as the largest

shareholder owns at least 35 % of the voting rights these variables have larger impact on audit

fees than for companies with a more diversified ownership structure. LOGNONFEE and

EXPOSURE display a significant negative relationship to ownership structure at the 35 %

cut-off level, this suggests that as the largest shareholder owns at least 35 % of the voting

rights the two variables will have less of an effect on the audit fees than for companies where

no shareholder owns at least 35 % of the voting rights. It interesting to note that as the cut-off

level changes from 50 % to 35 % the variable EXCOMP no longer displays a significant

relationship; this might be explained through the suggestions made by Fan & Wong (2005)

who stated that to gain credibility the largest shareholder might employ monitoring

mechanisms to assure the other shareholders that their best interests are being secured. As the

largest shareholder’s holds a larger percentage of the voting rights the need to employ

monitoring mechanisms to gain credibility might be accentuated.

6 The result from the modified secondary regression models is presented in Appendix 5

47

5 CONCLUDING REMARKS

The final chapter presents the conclusions drawn from the analysis (5.1), discusses the

limitations of the thesis (5.2) and suggests areas for further research regarding audit fee

determinants (5.3).

5.1 Conclusions

This thesis examines audit fee determinants for companies listed on the Nasdaq OMX

Stockholm Stock Exchange during 2001-2010 and whether these audit fee determinants differ

between ownership structures. Examining disclosed audit fees in the Swedish setting, which is

previously sparsely tested, contributes to the audit fee determinants area by providing further

evidence on the relationship between previously tested determinants and audit fees. The first

hypothesis can be accepted based on the support provided by the results as 7 of the 16

variables displayed a strong significant positive relationship. In consensus with previous

research a positive relationship to audit fees was established for size (Menon & Williams,

2001; Simunic, 1980), inherent risk (Stice, 1991; Taylor & Simon, 1999), foreign operations

as a proxy for complexity (Simon & Francis, 1988; Waresul Karim & Moizer, 1996),

Ownership structure (Nikkinen & Sahlström, 2004; Fan & Wong, 2005) and non-audit fees

(Firth, 1997; Palmrose, 1986b).

Furthermore, this study provides insight into earlier sparsely tested variables and finds a

significant relationship to audit fees for proportion of financial assets, the number of signing

auditor, executive compensation, impairment of intangibles and M&A. However, the results

could not confirm any significant relationships to audit fees for previously tested variables

free cash flow and audit fee premium indicating that the existence of an audit fee premium for

Big 4 firms is not present in the Swedish setting. The results from the test of number of audit

firms’ relation to audit fees showed no significant relationship. This is quite surprising as the

use of more than one audit firm should lead to a higher charged audit fee. The reason for the

result might indicate that as a company employs more than one audit firm the competition

increases thereby leaving the audit fee unchanged. The findings stand in contrast to previous

research by André et al. (2011) who found that joint audits lead to higher audit fees. However,

their study was performed on the French market which requires that a company employs at

least two audit firms which could lower the competition effect.

48

The second hypothesis was not supported by the results as the audit fee determinants for

different ownership structures do not differ substantially even though some minor

discrepancies exist when the majority owner controls at least 50% of the voting rights.

However, the robustness test did indicate more significant differences when the majority

owner controls at least 35% of the voting rights and the robustness tests revealed that the

second hypothesis could be accepted at a 35 % cut-off level. The robustness test also revealed

that it is predominantly business complexity and assurance that explains the variation in audit

fees determinants, indicating that these areas are more time consuming and/or complex

thereby increasing the audit price per unit. Furthermore, the robustness tests indicated that

size, non-audit fees and exposure explains a large portion of the variation in audit fees

suggesting that size is the predominant audit fee determinant which is in agreement with

previous research (e.g. Simon & Francis, 1988; Simunic, 1980).

An interesting result regarding the ownership structure test is that executive compensation

displays a significant difference depending on ownership structure when the majority owner

controls at least 50 % of the voting rights but as the majority owner controls at least 35 % the

executive compensation does not display a significant change in relationship to audit fees.

According to agency theory the relationship should be reversed (considering the separation of

ownership and control) as the agent’s and principals interests are more likely to align. The

explanation for this relationship can perhaps be found in the argument made by Fan & Wong

(2005). Fan & Wong (2005) states that as the controlling owner can appear less credible due

to the risk for self-serving behaviour he or she might implement a control mechanism to

assure the other owners that there interests are being protected thereby increasing the

monitoring need. This might be the case regarding executive compensation when the largest

shareholder controls at least 50 % of the voting rights. This area does need to be research

further to establish if Fan & Wong’s (2005) assumption holds for the Swedish setting.

5.2 Possible limitations

Although the study have been thoroughly considered and cautiously planned some limitations

have yet arisen. First of all, the study is conducted in a Swedish setting; therefore the data

utilized for the study only contains companies listed on the Nasdaq OMX Stockholm Stock

exchange. The utilization of listed companies on the Stockholm Stock exchange could affect

the generalizability of the study in a global context since the Swedish setting might contain

some country - specific characteristics e.g. legal liability regime (Choi et al., 2008). However,

49

since IFRS was introduced for EU members in 2005 all publically listed companies are

required to follow the reporting standards to increase the comparability.

Even though we are applying a large dataset, it does contain some limitations. The use of an

extended time period does have its benefits (e.g. more observations, results that can be

generalized to other years) but also its weaknesses. The applied time period could easily be

described as rather turbulent with the introduction of new standards and financial crises. For

example, Kim et al. (2012) found that the introduction of IFRS put upwards pressure on audit

fees during the years of implementation.

Due to limitations in the dataset we encountered problem with absence of variables. For

example, the Swedish setting requires the disclosure of audit partner identity for an

engagement (Zerni, 2012). This could potentially be an explanatory variable for audit fees

which have been omitted from this study due to absence in the database. Other possible

explanatory variables not tested could be the amount of internal auditing or level of auditor

expertise. Further research could possibly try to implement these variables.

5.3 Suggestions for further studies

We suggest that other studies on audit fees tries to incorporate even more variables in their

models to find further consensus about audit fee determinants. For example, Felix, Gramling

& Maletta (2001) finds that internal audit contribution is a significant determinant of audit

fees. This study also provided initial insight into the audit fee determinants for different

ownership structures; this area does however require more research where other ownership

structures are tested against audit fees.

Our finding suggests that executive compensation is a significant factor in determining audit

fees for companies with a majority owner controlling at least 50% of the voting rights. The

executive compensation is a larger explanatory factor for audit fees in companies with a

majority owner controlling at least 50% of the voting rights then in companies with a more

separated ownership structure. These finding are quite interesting and further studies

investigating these results are needed to understand the underlying reasons. Furthermore,

deeper studies investigating the relationship between audit fees and non-audit fees can

contribute to the research literature by identifying the possible absence of knowledge spillover

and the proposed “Chinese wall” that audit firms raise between consulting and auditing

engagements.

I

REFERENCES

Books

Bryman, A., & Bell, E. (2011). Business Research Methods (3rd edition.). New York, United

States of America: Oxford University Press Inc.

Lind, D. A., Marchal, W. G., & Wathen, S. A. (2010). Statistical Techniques in Business and

Economics, (14th Ed.). New York, United States of America: McGraw-Hill/Irwin.

Power, M. (1999). The audit society: rituals of verification. 2010 Edition. Oxford: Oxford

Univ. Press.

Conference papers

Esteves, J. (2006). Establishing the relationship between enterprise systems benefits, business

complexity, and business alignment in SMEs. In: European and Mediterranean

Conference on Information Systems (EMCIS), Costa Blanca, Alicante, Spain

Journals

André, P., Broye, G., Pong, C., & Schatt, A. (2010). Audit Fees, Big Four Premium and

Institutional Settings: The Devil is in the Details! (SSRN Scholarly Paper No. ID

1554842). Rochester, NY: Social Science Research Network.

Angwin, D. (2007). Motive Archetypes in Mergers and Acquisitions (M&A): The

Implications of a Configurational Approach to Performance. Advances in Mergers &

Acquisitions. 6, 77–105.

Bens, D. A., & Johnston, R. (2009). Accounting Discretion: Use or Abuse? An Analysis of

Restructuring Charges Surrounding Regulator Action. Contemporary Accounting

Research. 26(3), 673–699.

Bens, D. A., Heltzer, W., & Segal, B. (2011). The Information Content of Goodwill

Impairments and SFAS 142. Journal of Accounting, Auditing & Finance. 26(3), 527–

555.

Brammer, S., & Millington, A. (2005). Corporate Reputation and Philanthropy: An Empirical

Analysis. Journal of Business Ethics. 61(1), 29–44.

Brammer, S., & Pavelin, S. (2006). Corporate Reputation and Social Performance: The

Importance of Fit. Journal of Management Studies. 43(3), 435–455.

Caneghem, T. V. (2010a). Audit pricing and the Big4 fee premium: evidence from Belgium.

Managerial Auditing Journal. 25(2), 122–139.

II

Cartwright, S., & Schoenberg, R. (2006). Thirty Years of Mergers and Acquisitions Research:

Recent Advances and Future Opportunities. British Journal of Management. 17(1), 1-

5.

Chng, D. H. M., Rodgers, M. S., Shih, E., & Song, X.-B. (2012). When does incentive

compensation motivate managerial behaviors? An experimental investigation of the fit

between incentive compensation, executive core self-evaluation, and firm

performance. Strategic Management Journal. 33(12), 1343–1362.

Choi, J.-H., Kim, J.-B., Liu, X., & Simunic, D. A. (2008). Audit Pricing, Legal Liability

Regimes, and Big 4 Premiums: Theory and Cross-country Evidence*. Contemporary

Accounting Research. 25(1), 55–99.

Cobbin, P. E. (2002). International Dimensions of the Audit Fee Determinants Literature.

International Journal of Auditing. 6(1), 53–77.

Craninckx, K., & Huyghebaert, N. (2011). Can Stock Markets Predict M&A Failure? A Study

of European Transactions in the Fifth Takeover Wave. European Financial

Management. 17(1), 9–45.

Dechow, P. M., Richardson, S. A., & Sloan, R. G. (2008a). The Persistence and Pricing of the

Cash Component of Earnings. Journal of Accounting Research. 46(3), 537–566.

Ebrahim, A. (2010). Audit fee premium and auditor change: the effect of Sarbanes-Oxley Act.

Managerial Auditing Journal. 25(2), 102–121.

Fan, J. P. H., & Wong, T. J. (2005). Do External Auditors Perform a Corporate Governance

Role in Emerging Markets? Evidence from East Asia. Journal of Accounting

Research. 43(1), 35–72.

Felix Jr., W. L., Gramling, A. A., & Maletta, M. J. (2001). The Contribution of Internal Audit

as a Determinant of External Audit Fees and Factors Influencing This Contribution.

Journal of Accounting Research. 39(3), 513–534.

Firth, M. (1997). The Provision of Non-Audit Services and the Pricing of Audit Fees. Journal

of Business Finance & Accounting. 24(3/4), 511–525.

Fleming, G., Heaney, R., & McCosker, R. (2005). Agency costs and ownership structure in

Australia. Pacific-Basin Finance Journal. 13(1), 29–52.

Fox, M. A., & Hamilton, R. T. (1994). Ownership and Diversification: Agency Theory or

Stewardship Theory. Journal of Management Studies. 31(1), 69–81.

Francis, J. R. (1994). Auditing, hermeneutics, and subjectivity. Accounting, Organizations

and Society. 19(3), 235–269.

III

Francis, J. R. (2004). What do we know about audit quality? The British Accounting Review.

36(4), 345–368.

Francis, J. R., & Simon, D. T. (1987). A Test of Audit Pricing in the Small-Client Segment of

the U.S. Audit Market. Accounting Review. 62(1), 145–157.

Francis, J. R., & Wang, D. (2005). Impact of the SEC’s Public Fee Disclosure Requirement

on Subsequent Period Fees and Implications for Market Efficiency. Auditing: A

Journal of Practice & Theory. 24(1), 145–160.

Gregoire, T. G., & Schabenberger, O. (1999). Sampling-Skewed Biological Populations:

Behavior of Confidence Intervals for the Population Total. Ecology. 80(3), 1056–1065

Gul, F. A., Tsui, J., & Chen, C. (1998). Agency Costs and Audit Pricing: Evidence on

Discretionary Accruals (SSRN Scholarly Paper No. ID 62750). Rochester, NY: Social

Science Research Network.

Haapamäki, E., Järvinen, T., Niemi, L. & Zerni, M. (2011). Do Joint Audits Offer Value for

Money? Abnormal Accruals, Earnings Conservatism, and Auditor Remuneration in a

Setting of Voluntary Joint Audits. Working Paper, University of Vaasa and Aalto

University.

Hay, D. C., Knechel, W. R., & Wong, N. (2006). Audit Fees: A Meta-analysis of the Effect of

Supply and Demand Attributes. Contemporary Accounting Research. 23(1), 141–191.

Hernandez, M. (2012). Toward an Understanding of the Psychology of Stewardship. Academy

of Management Review. 37(2), 172–193.

Jensen, M. (1986). Agency cost of free cash flow, corporate finance, and takeovers. Corporate

Finance, and Takeovers. American Economic Review. 76(2)

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency

costs and ownership structure. Journal of Financial Economics. 3(4), 305–360.

Kim, J.B., Liu, X. & Zheng, L. (2012). The Impact of Mandatory IFRS Adoption on Audit

Fees: Theory and Evidence. Accounting Review, 87(6), 2061–2094.

Krishnan, G. V., & Yu, W. (2011). Further evidence on knowledge spillover and the joint

determination of audit and non-audit fees. Managerial Auditing Journal. 26(3), 230–

247.

Lee, B. (2002). Professional socialisation, commercial pressures and Junior staff’s time-

pressured irregular Auditing- A contextual interpretation. The British Accounting

Review. 34(4), 315–333.

IV

Maletta, M. J. (1993). An Examination of Auditors’ Decisions to Use Internal Auditors as

Assistants: The Effect of Inherent Risk. Contemporary Accounting Research. 9(2),

508–525.

Meglio, O., & Risberg, A. (2010). Mergers and acquisitions—Time for a methodological

rejuvenation of the field? Scandinavian Journal of Management. 26(1), 87–95.

Menon, K., & Williams, D. D. (2001). Long-Term Trends in Audit Fees. Auditing: A Journal

of Practice & Theory. 20(1), 115–136.

Mitnick, B. M. (1973). Fiduciary Rationality and Public Policy: The Theory of Agency and

Some Consequences. New Orleans, LA. IMU Serials.

Moehrle, S. R. (2002). Do Firms Use Restructuring Charge Reversals to Meet Earnings

Targets? Accounting Review. 77(2), 397–413.

Morse, A., Nanda, V., & Seru, A. (2011). Are Incentive Contracts Rigged by Powerful CEOs?

Journal of Finance. 66(5), 1779–1821.

Mustapha, M., & Ahmad, A. C. (2011). Agency theory and managerial ownership: evidence

from Malaysia. Managerial Auditing Journal. 26(5), 419–436.

Nguyen, H. T., Yung, K., & Sun, Q. (2012). Motives for Mergers and Acquisitions: Ex-Post

Market Evidence from the US. Journal of Business Finance & Accounting. 39(9/10),

1357–1375.

Nikkinen, J., & Sahlström, P. (2004). Does Agency Theory Provide a General Framework for

Audit Pricing? International Journal of Auditing. 8(3), 253–262.

Nyberg, A. J., Fulmer, I. S., Gerhart, B., & Carpenter, M. A. (2010). Agency Theory

Revisited: Ceo Return and Shareholder Interest Alignment. Academy of Management

Journal. 53(5), 1029–1049.

Palmrose, Z.-V. (1986a). Audit Fees and Auditor Size: Further Evidence. Journal of

Accounting Research. 24(1), 97–110.

Palmrose, Z.-V. (1986b). The Effect of Nonaudit Services on the Pricing of Audit Services:

Further Evidence. Journal of Accounting Research. 24(2), 405–411.

Peek, E. (2004). The Use of Discretionary Provisions in Earnings Management: Evidence

from The Netherlands. Journal of International Accounting Research. 3(2), 27–43.

Pentland, B. T. (1993). Getting comfortable with the numbers: Auditing and the micro-

production of macro-order. Accounting, Organizations and Society. 18(7–8), 605–620.

Pentland, B. T. (2000). Will auditors take over the world? Program, technique and the

verification of everything. Accounting, Organizations and Society. 25(3), 307–312.

V

Peterson, K. (2012). Accounting complexities, misreporting, and the consequences of

misreporting. Review of Accounting Studies. 17(1), 72-95

Power, M. (2003). Auditing and the production of legitimacy. Accounting, Organizations and

Society. 28(4), 379–394.

Quick, R. (2012). EC Green Paper Proposals and Audit Quality. Accounting in Europe. 9(1),

17–38.

Reitman, D. (1993). Stock Options and the Strategic Use of Managerial Incentives. The

American Economic Review. 83(3), 513–524.

Roberts, P. W., & Dowling, G. R. (2002). Corporate reputation and sustained superior

financial performance. Strategic Management Journal. 23(12), 1077–1093.

Ross, S. A. (1973). The Economic Theory of Agency: The Principal’s Problem. The American

Economic Review. 63(2), 134–139.

Schilt, H. M., & Perler, J. (2010a). Financial Shenanigans - How to detect accounting

gimmicks & Fraud in financial reports (3rd Ed.). United States: McGraw-Hill.

Simon, D. T. (1985). The Audit Services Market: Additional Empirical Evidence. Auditing: A

Journal of Practice & Theory. 5(1), 71.

Simon, D. T., & Francis, J. R. (1988). The Effects of Auditor Change on Audit Fees: Tests of

Price Cutting and Price Recovery. Accounting Review. 63(2), 255–269.

Simunic, D. A. (1980). The Pricing of Audit Services: Theory and Evidence. Journal of

Accounting Research. 18(1), 161–190.

Srinidhi, B. N., & Gul, F. A. (2007). The Differential Effects of Auditors’ Nonaudit and Audit

Fees on Accrual Quality. Contemporary Accounting Research. 24(2), 595–629.

Stanley, J. D. (2011). Is the Audit Fee Disclosure a Leading Indicator of Clients’ Business

Risk? Auditing: A Journal of Practice & Theory. 30(3), 157–179.

Stice, J. D. (1991). Using Financial and Market Information to Identify Pre-Engagement

Factors Associated with Lawsuits Against Auditors. Accounting Review. 66(3), 516–

533.

Suddaby, R., Gendron, Y., & Lam, H. (2009). The organizational context of professionalism

in accounting. Accounting, Organizations and Society. 34(3–4), 409–427.

Sundgren, S., & Svanström, T. (2013). Audit office size, audit quality and audit pricing:

evidence from small- and medium-sized enterprises. Accounting and Business

Research. 43(1), 31–55.

VI

Taylor, M. H., & Simon, D. T. (1999). Determinants of audit fees: the importance of

litigation, disclosure, and regulatory burdens in audit engagements in 20 countries. The

International Journal of Accounting. 34(3), 375–388.

Thinggaard, F., & Kiertzner, L. (2008). Determinants of Audit Fees: Evidence from a Small

Capital Market with a Joint Audit Requirement. International Journal of Auditing.

12(2), 141–158.

Wallace, W.A. (2004). The Economic Role of the Audit in Free and Regulated Markets: A

Look Back and a Look Forward. Research in Accounting Regulation. 17, 267–98

Waresul Karim, A. K. M., & Moizer, P. (1996). Determinants of audit fees in Bangladesh.

The International Journal of Accounting. 31(4), 497–509.

Zerni, M. (2012). Audit Partner Specialization and Audit Fees: Some Evidence from

Sweden*. Contemporary Accounting Research. 29(1), 312–340.

Legal statements and standards

Green Paper, 2010. Audit Policy: Lesson from the Crisis, European Commission

IAS 36, 2004. International accounting standards: Impairment of Assets, IASCF Publications

Dept.

IAS 37, 2001. International accounting standards: Provisions, Contingent Liabilities and

Contingent Assets, IASCF Publications Dept.

ISA 200, 2009. Overall objectives of the independent auditor and the conduct of an audit in

accordance with international standards on auditing, International Federation of

Accountants

SFS 1995:1554. Annual Accounts Act. Sweden, Ministry of Justice

SFS 2005:551. The Companies Act. Sweden, Ministry of Justice

VII

APPENDIX

Appendix 1 – Test of Variance Inflation Factor

Primary regression model for audit fee determinants

Model Unstandardized

Coefficients

Standardized

Coefficients

t Sig. Collinearity

Statistics

B Std. Error Beta Tolerance VIF

1

(Constant) 2,081 ,080 26,120 ,000

PREMIUM -,031 ,041 -,006 -,757 ,449 ,884 1,131

LOGNONFEE ,200 ,013 ,223 15,461 ,000 ,290 3,454

EXPOSURE ,007 ,006 ,014 1,237 ,216 ,446 2,244

MA ,086 ,034 ,020 2,519 ,012 ,941 1,063

EXCOMP ,057 ,023 ,020 2,523 ,012 ,965 1,036

LOGTOTASS ,509 ,013 ,684 38,908 ,000 ,195 5,121

FOROP ,346 ,036 ,083 9,584 ,000 ,811 1,234

FINANCE -,192 ,080 -,023 -2,408 ,016 ,663 1,507

IRISK ,957 ,075 ,113 12,732 ,000 ,770 1,298

IMPAIR 1,846 ,324 ,047 5,695 ,000 ,867 1,153

ACCRUAL -,001 ,001 -,015 -1,851 ,064 ,873 1,146

PROV -,006 ,006 -,007 -,918 ,359 ,990 1,010

FCFA ,017 ,016 ,009 1,053 ,292 ,743 1,346

OWNERS ,098 ,058 ,014 1,696 ,090 ,897 1,115

NUMFIRM -,049 ,047 -,009 -1,041 ,298 ,865 1,156

NUMAUD -,062 ,025 -,020 -2,463 ,014 ,925 1,081

Appendix 2 – Primary regression model for audit fee determinants

Coefficientsa

Model Unstandardized Coefficients Standardized

Coefficients

t Sig.

B Std. Error Beta

1

(Constant) 2,081 ,080 26,120 ,000

PREMIUM -,031 ,041 -,006 -,757 ,449

LOGNONFEE ,200 ,013 ,223 15,461 ,000

EXPOSURE ,007 ,006 ,014 1,237 ,216

MA ,086 ,034 ,020 2,519 ,012

EXCOMP ,057 ,023 ,020 2,523 ,012

VIII

LOGTOTASS ,509 ,013 ,684 38,908 ,000

FOROP ,346 ,036 ,083 9,584 ,000

FINANCE -,192 ,080 -,023 -2,408 ,016

IRISK ,957 ,075 ,113 12,732 ,000

IMPAIR 1,846 ,324 ,047 5,695 ,000

ACCRUAL -,001 ,001 -,015 -1,851 ,064

PROV -,006 ,006 -,007 -,918 ,359

FCFA ,017 ,016 ,009 1,053 ,292

OWNERS ,098 ,058 ,014 1,696 ,090

NUMFIRM -,049 ,047 -,009 -1,041 ,298

NUMAUD -,062 ,025 -,020 -2,463 ,014

a. Dependent Variable: LOGFEE

Model Summary

Model R R Square Adjusted R

Square

Std. Error of

the Estimate

Change Statistics

R Square

Change

F Change df1 df2 Sig. F

Change

1 ,937a ,877 ,876 ,502 ,877 909,224 16 2035 ,000

a. Predictors: (Constant), NUMAUD, PROV, ACCRUAL, EXCOMP, FINANCE, IMPAIR, PREMIUM, FOROP, MA, OWNERS,

NUMFIRM, EXPOSURE, FCFA, IRISK, LOGNONFEE, LOGTOTASS

Appendix 3 – Secondary regression model for audit fee determinants – adjusted for

ownership structure

Secondary regression model for audit fee determinants

Adjusted for ownership structure

Model Unstandardized Coefficients Standardized

Coefficients

t Sig.

B Std. Error Beta

(Constant) 2,093 ,079 26,462 ,000

PREMIUM -,015 ,046 -,003 -,333 ,739

LOGNONFEE ,206 ,014 ,230 14,570 ,000

EXPOSURE ,011 ,007 ,021 1,597 ,111

MA ,075 ,037 ,017 2,032 ,042

EXCOMP ,033 ,025 ,011 1,299 ,194

LOGTOTASS ,508 ,014 ,682 35,914 ,000

FOROP ,283 ,040 ,068 7,146 ,000

FINANCE -,231 ,086 -,028 -2,683 ,007

IRISK ,950 ,082 ,112 11,556 ,000

IX

IMPAIR 1,685 ,346 ,043 4,877 ,000

ACCRUAL -,001 ,001 -,016 -1,854 ,064

PROV -,005 ,006 -,007 -,839 ,402

FCFA ,021 ,017 ,012 1,238 ,216

NUMFIRM -,044 ,054 -,008 -,809 ,419

NUMAUD -,047 ,028 -,015 -1,670 ,095

PREMIUM * OWNCON -,016 ,105 -,004 -,155 ,877

LOGNONFEE * OWNCON -,057 ,033 -,108 -1,728 ,084

EXPOSURE * OWNCON -,020 ,014 -,034 -1,424 ,155

MA * OWNCON ,061 ,103 ,005 ,594 ,552

EXCOMP * OWNCON ,129 ,059 ,024 2,188 ,029

LOGTOTASS * OWNCON ,029 ,033 ,062 ,886 ,376

FOROP * OWNCON ,373 ,102 ,056 3,667 ,000

FINANCE * OWNCON ,212 ,168 ,014 1,267 ,205

IRISK * OWNCON ,146 ,171 ,016 ,857 ,392

IMPAIR * OWNCON 1,410 ,989 ,013 1,426 ,154

ACCRUAL * OWNCON ,002 ,002 ,008 ,895 ,371

PROV * OWNCON ,058 2,334 ,000 ,025 ,980

FCFA * OWNCON -,032 ,047 -,007 -,681 ,496

NUMFIRM * OWNCON ,006 ,108 ,001 ,057 ,954

NUMAUD * OWNCON -,094 ,064 -,017 -1,461 ,144

Model Summary

Model R R Square Adjusted R

Square

Std. Error of

the Estimate

Change Statistics

R Square

Change

F Change df1 df2 Sig. F

Change

1 ,938a ,879 ,877 ,50006 ,879 489,120 30 2021 ,000

X

Appendix 4 – Primary regression model modified by exclusion of variables

Primary regression model excluding LOGTOTASS

Model R R Square Adjusted R

Square

Std. Error of the

Estimate

1 ,887a ,786 ,784 ,662

a. Predictors: (Constant), NUMAUD, PROV, ACCRUAL, EXCOMP,

FINANCE, IMPAIR, PREMIUM, FOROP, MA, OWNERS, NUMFIRM,

EXPOSURE, FCFA, IRISK, LOGNONFEE

Primary regression model excluding LOGNONFEE

Model R R Square Adjusted R

Square

Std. Error of the

Estimate

1 ,929a ,863 ,862 ,529

a. Predictors: (Constant), LOGTOTASS, PROV, IRISK, EXCOMP,

ACCRUAL, MA, NUMFIRM, IMPAIR, OWNERS, NUMAUD,

PREMIUM, FOROP, FCFA, FINANCE, EXPOSURE

Primary regression model excluding EXPOSURE

Model R R Square Adjusted R

Square

Std. Error of the

Estimate

1 ,910a ,855 ,852 ,502

a. Predictors: (Constant), LOGNONFEE, PROV, IRISK, ACCRUAL,

MA, EXCOMP, NUMFIRM, IMPAIR, OWNERS, NUMAUD, PREMIUM,

FOROP, FCFA, FINANCE, LOGTOTASS

Primary regression model excluding LOGTOTASS,

LOGNONFEE and EXPOSURE

Model R R Square Adjusted R

Square

Std. Error of the

Estimate

1 ,621a ,386 ,382 1,121

a. Predictors: (Constant), FCFA, EXCOMP, NUMFIRM, PROV, IRISK,

MA, FOROP, OWNERS, NUMAUD, PREMIUM, IMPAIR, ACCRUAL,

FINANCE

XI

Appendix 5 – Modified secondary regression model – Cut-off level 35 %

Secondary regression model, ownership cut-off level 35 %

Model R R Square Adjusted R

Square

Std. Error of

the Estimate

Change Statistics

R Square

Change

F Change df1 df2 Sig. F

Change

1 ,938a ,880 ,878 ,49816 ,880 493,377 30 2021 ,000

a. Predictors: (Constant), NUMAUD * OWNCON, LOGTOTASS , PROV, PROV * OWNCON, EXCOMP, ACCRUAL, MA, IMPAIR *

OWNCON, NUMFIRM * OWNCON, IRISK, PREMIUM, FCFA * OWNCON, FOROP, FINANCE * OWNCON, IMPAIR , MA *

OWNCON, NUMAUD, FCFA , EXCOMP * OWNCON, FINANCE , ACCRUAL * OWNCON, FOROP * OWNCON, EXPOSURE,

NUMFIRM, IRISK * OWNCON, LOGNONFEE, EXPOSURE * OWNCON, PREMIUM * OWNCON, LOGNONFEE * OWNCON,

LOGTOTASS * OWNCON

Coefficientsa for secondary regression model, ownership cut-off level 35 %

Model Unstandardized Coefficients Standardized

Coefficients

t Sig.

B Std. Error Beta

1

(Constant) 2,073 ,080 25,839 ,000

PREMIUM ,009 ,054 ,002 ,174 ,862

LOGNONFEE ,222 ,016 ,248 14,145 ,000

EXPOSURE ,018 ,007 ,035 2,451 ,014

MA ,099 ,041 ,023 2,407 ,016

EXCOMP ,051 ,028 ,018 1,813 ,070

LOGTOTASS ,486 ,016 ,653 31,322 ,000

FOROP ,246 ,044 ,059 5,585 ,000

FINANCE -,327 ,092 -,039 -3,541 ,000

IRISK 1,009 ,090 ,119 11,247 ,000

IMPAIR 1,528 ,390 ,039 3,916 ,000

ACCRUAL -,002 ,001 -,024 -2,358 ,018

PROV -,005 ,006 -,006 -,800 ,424

FCFA ,026 ,019 ,014 1,368 ,172

NUMFIRM 4,953E-005 ,067 ,000 ,001 ,999

NUMAUD -,048 ,031 -,015 -1,529 ,127

PREMIUM * OWNCON -,039 ,080 -,013 -,482 ,630

LOGNONFEE * OWNCON -,076 ,026 -,180 -2,883 ,004

EXPOSURE * OWNCON -,026 ,012 -,055 -2,139 ,033

MA * OWNCON -,051 ,073 -,007 -,698 ,485

EXCOMP * OWNCON ,026 ,047 ,006 ,538 ,591

XII

LOGTOTASS * OWNCON ,077 ,026 ,201 2,931 ,003

FOROP * OWNCON ,298 ,078 ,059 3,847 ,000

FINANCE * OWNCON ,393 ,135 ,034 2,914 ,004

IRISK * OWNCON -,144 ,137 -,019 -1,047 ,295

IMPAIR * OWNCON 1,030 ,698 ,015 1,477 ,140

ACCRUAL * OWNCON ,002 ,001 ,019 1,774 ,076

PROV * OWNCON ,825 ,757 ,008 1,089 ,276

FCFA * OWNCON -,049 ,036 -,015 -1,386 ,166

NUMFIRM * OWNCON -,062 ,094 -,008 -,657 ,511

NUMAUD * OWNCON -,035 ,052 -,008 -,664 ,506

a. Dependent Variable: LOGFEE